ivan-workspace/First_iteration.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "import datetime as dt\n",
      "from collections import defaultdict, Counter, namedtuple\n",
      "from itertools import product, combinations, permutations\n",
      "import json\n",
      "import os\n",
      "import pickle\n",
      "import warnings\n",
      "\n",
      "from IPython.display import HTML, display, set_matplotlib_formats\n",
      "import numpy as np\n",
      "import pandas as pd\n",
      "import matplotlib.pyplot as plt\n",
      "import seaborn as sns\n",
      "\n",
      "from sqlalchemy import create_engine\n",
      "from tqdm.auto import tqdm\n",
      "\n",
      "# INITIALIZING\n",
      "# warnings\n",
      "warnings.filterwarnings('ignore')\n",
      "# pictures\n",
      "sns.set()\n",
      "plt.style.use('seaborn-bright')\n",
      "plt.rcParams['figure.figsize'] = 10, 6\n",
      "# pandas\n",
      "pd.set_option('display.max_columns', 150)\n",
      "pd.set_option('display.max_rows', 100)\n",
      "tqdm.pandas()\n",
      "\n",
      "# Autoreload\n",
      "get_ipython().magic(u\"%reload_ext autoreload\")\n",
      "get_ipython().magic(u\"%autoreload 2\")\n",
      "get_ipython().magic(u\"%load_ext line_profiler\")\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(_IMPORTS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import re\n",
    "\n",
    "from glob import glob\n",
    "\n",
    "from typing import List, Optional"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "from sklearn.metrics import roc_auc_score\n",
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction.text import TfidfVectorizer\n",
    "from sklearn.feature_extraction import text"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "import torch\n",
    "from transformers import AutoTokenizer, AutoModelForMaskedLM, AutoConfig"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Sampling on patients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "PATH = \"../data/raw/\"\n",
    "PATH_PROCESSED = \"../data/processed/\"\n",
    "PATH_DATASETS = \"../data/train/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "SAMPLE_SIZE = 1000\n",
    "RANDOM_SEED = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    30761\n",
       "1    15759\n",
       "Name: EXPIRE_FLAG, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "patients = pd.read_csv(os.path.join(PATH, \"PATIENTS.csv.gz\"))\n",
    "patients.EXPIRE_FLAG.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "46520 unique patients in 46520 rows\n"
     ]
    }
   ],
   "source": [
    "#ensuring every patient is unique\n",
    "print(f\"{patients.SUBJECT_ID.nunique()} unique patients in {len(patients)} rows\")\n",
    "#sampling random patients\n",
    "patients_sample = patients#.sample(n=SAMPLE_SIZE, random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "SAMPLE_IDS = set(patients_sample.SUBJECT_ID)\n",
    "\n",
    "with open(os.path.join(PATH_PROCESSED, \"SAMPLE_IDS.json\"), 'w') as f:\n",
    "    json.dump({'ids': list(SAMPLE_IDS)}, f)\n",
    "    \n",
    "with open(os.path.join(PATH_PROCESSED, \"SAMPLE_IDS.json\"), 'r') as f:\n",
    "    SAMPLE_IDS = set(json.load(f)['ids'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "patients_sample = patients[patients.SUBJECT_ID.isin(SAMPLE_IDS)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ROW_ID</th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>GENDER</th>\n",
       "      <th>DOB</th>\n",
       "      <th>DOD</th>\n",
       "      <th>DOD_HOSP</th>\n",
       "      <th>DOD_SSN</th>\n",
       "      <th>EXPIRE_FLAG</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>234</td>\n",
       "      <td>249</td>\n",
       "      <td>F</td>\n",
       "      <td>2075-03-13 00:00:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>235</td>\n",
       "      <td>250</td>\n",
       "      <td>F</td>\n",
       "      <td>2164-12-27 00:00:00</td>\n",
       "      <td>2188-11-22 00:00:00</td>\n",
       "      <td>2188-11-22 00:00:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>236</td>\n",
       "      <td>251</td>\n",
       "      <td>M</td>\n",
       "      <td>2090-03-15 00:00:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>237</td>\n",
       "      <td>252</td>\n",
       "      <td>M</td>\n",
       "      <td>2078-03-06 00:00:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>238</td>\n",
       "      <td>253</td>\n",
       "      <td>F</td>\n",
       "      <td>2089-11-26 00:00:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ROW_ID  SUBJECT_ID GENDER                  DOB                  DOD  \\\n",
       "0     234         249      F  2075-03-13 00:00:00                  NaN   \n",
       "1     235         250      F  2164-12-27 00:00:00  2188-11-22 00:00:00   \n",
       "2     236         251      M  2090-03-15 00:00:00                  NaN   \n",
       "3     237         252      M  2078-03-06 00:00:00                  NaN   \n",
       "4     238         253      F  2089-11-26 00:00:00                  NaN   \n",
       "\n",
       "              DOD_HOSP DOD_SSN  EXPIRE_FLAG  \n",
       "0                  NaN     NaN            0  \n",
       "1  2188-11-22 00:00:00     NaN            1  \n",
       "2                  NaN     NaN            0  \n",
       "3                  NaN     NaN            0  \n",
       "4                  NaN     NaN            0  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "patients_sample.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "DECEASED_TO_DATE = patients_sample[patients_sample.EXPIRE_FLAG == 1]\\\n",
    "    .set_index('SUBJECT_ID').DOD.map(lambda x: pd.to_datetime(x).date()).to_dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_data_for_sample(sample_ids: set, \n",
    "                        file_name: str, \n",
    "                        chunksize: int = 10_000) -> pd.DataFrame:\n",
    "    \"\"\"Get the data only relevant for the sample.\"\"\"\n",
    "    full_path = os.path.join(PATH, file_name)\n",
    "    iterator = pd.read_csv(full_path, iterator=True, chunksize=chunksize)\n",
    "    return pd.concat([chunk[chunk.SUBJECT_ID.isin(SAMPLE_IDS)] for chunk in tqdm(iterator)])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build features"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "All features in feature_prepocessed form are features with columns ['SUBJECT_ID', 'FEATURE_NAME', 'DATE', 'VALUE], which can be later used for any of the aggregations we'd like."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a2e4cbe3f9964d758e51057707b47409",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "admissions = get_data_for_sample(SAMPLE_IDS, \"ADMISSIONS.csv.gz\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "admissions['ADMITTIME'] = pd.to_datetime(admissions.ADMITTIME).dt.date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "aef770ce29a54c3d8297b0c356c52ac3",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "diagnoses = get_data_for_sample(SAMPLE_IDS, \"DIAGNOSES_ICD.csv.gz\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "08c053db91e0496eab05152f51c934f8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "lab_results = get_data_for_sample(SAMPLE_IDS, \"LABEVENTS.csv.gz\", chunksize=100_000)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2a58c30243a946adbea7af94688c8255",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "meds = get_data_for_sample(SAMPLE_IDS, \"PRESCRIPTIONS.csv.gz\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Diagnoses"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "diagnoses['ICD9_CODE'] = \"ICD9_\" + diagnoses['ICD9_CODE']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "adm_cols = ['SUBJECT_ID', 'HADM_ID', 'ADMITTIME']\n",
    "diagnoses = diagnoses.merge(admissions[adm_cols], on=['SUBJECT_ID', 'HADM_ID'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "dropper = ['ROW_ID', 'SEQ_NUM', 'HADM_ID']\n",
    "renamer = {'ICD9_CODE': 'FEATURE_NAME', 'ADMITTIME': 'DATE'}\n",
    "diag_preprocessed = diagnoses.drop(columns=dropper).rename(columns=renamer)\n",
    "diag_preprocessed['VALUE'] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>FEATURE_NAME</th>\n",
       "      <th>DATE</th>\n",
       "      <th>VALUE</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>109</td>\n",
       "      <td>ICD9_40301</td>\n",
       "      <td>2141-09-18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>109</td>\n",
       "      <td>ICD9_486</td>\n",
       "      <td>2141-09-18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>109</td>\n",
       "      <td>ICD9_58281</td>\n",
       "      <td>2141-09-18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>109</td>\n",
       "      <td>ICD9_5855</td>\n",
       "      <td>2141-09-18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>109</td>\n",
       "      <td>ICD9_4254</td>\n",
       "      <td>2141-09-18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   SUBJECT_ID FEATURE_NAME        DATE  VALUE\n",
       "0         109   ICD9_40301  2141-09-18      1\n",
       "1         109     ICD9_486  2141-09-18      1\n",
       "2         109   ICD9_58281  2141-09-18      1\n",
       "3         109    ICD9_5855  2141-09-18      1\n",
       "4         109    ICD9_4254  2141-09-18      1"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "diag_preprocessed.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Labs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "lab_results['DATE'] = pd.to_datetime(lab_results['CHARTTIME']).dt.date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "lab_results['FEATURE_NAME'] = \"LAB_\" + lab_results['ITEMID'].astype(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "dropper = ['ROW_ID', 'HADM_ID', 'VALUE', 'VALUEUOM', 'FLAG', 'ITEMID', 'CHARTTIME']\n",
    "renamer = {'VALUENUM': 'VALUE'}\n",
    "lab_preprocessed = lab_results.drop(columns=dropper).rename(columns=renamer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>VALUE</th>\n",
       "      <th>DATE</th>\n",
       "      <th>FEATURE_NAME</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>3</td>\n",
       "      <td>7.39</td>\n",
       "      <td>2101-10-12</td>\n",
       "      <td>LAB_50820</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2101-10-12</td>\n",
       "      <td>LAB_50800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>3</td>\n",
       "      <td>-1.00</td>\n",
       "      <td>2101-10-12</td>\n",
       "      <td>LAB_50802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>22.00</td>\n",
       "      <td>2101-10-12</td>\n",
       "      <td>LAB_50804</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3</td>\n",
       "      <td>0.93</td>\n",
       "      <td>2101-10-12</td>\n",
       "      <td>LAB_50808</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   SUBJECT_ID  VALUE        DATE FEATURE_NAME\n",
       "0           3   7.39  2101-10-12    LAB_50820\n",
       "1           3    NaN  2101-10-12    LAB_50800\n",
       "2           3  -1.00  2101-10-12    LAB_50802\n",
       "3           3  22.00  2101-10-12    LAB_50804\n",
       "4           3   0.93  2101-10-12    LAB_50808"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lab_preprocessed.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Meds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "meds = meds[meds.ENDDATE.notna()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "meds['DATE'] = pd.to_datetime(meds['ENDDATE']).dt.date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def find_mean_dose(dose: str) -> float:\n",
    "    if pd.isnull(dose):\n",
    "        return 0\n",
    "    try:\n",
    "        cleaned = re.sub(r'[A-Za-z,>< ]', '', dose)\n",
    "        parts = cleaned.split('-')\n",
    "        return np.array(parts).astype(float).mean()\n",
    "    except:\n",
    "        print(dose)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "50/500\n",
      "250/50\n",
      "500//50\n",
      "1%\n",
      "1%\n",
      "1%\n",
      "1.25%\n",
      "800/160\n",
      "-0.5-2\n",
      "0.3%\n",
      "15-30 mmol and Call H.O.\n",
      "About-CM1000\n",
      "one\n",
      "About-CM50\n",
      "one\n",
      "one\n",
      "500/50\n",
      "4 GM and Call H.O.\n",
      "standard\n",
      "2/2.5\n",
      "12-\n",
      "4 GM and Call H.O.\n",
      "'5\n",
      "0.625/2.5\n",
      "0.05%\n",
      "0.05%\n",
      "-15-30\n",
      "2mcg/kg\n",
      "\\12.5\n",
      "15-30 mmol and Call H.O.\n",
      "16; give 1/2 dose when NPO\n",
      "7.5mg (3\n",
      "7.5mg (3\n",
      "7.5mg (3\n",
      "7.5mg (3\n",
      "1%\n",
      "Hold Dose\n",
      "5%\n",
      "1.25/3\n",
      "-1000\n",
      "-500\n",
      "1.36/6\n",
      "1.26/6\n",
      "1%\n",
      "50--100\n",
      "40-\n",
      ": 5-10\n",
      "0.63/3\n",
      "1-\n",
      "0.63/3\n",
      "12.5--25\n",
      "20-\n",
      "0.5mg/5\n",
      "50-\n",
      "0.5--2\n",
      "1.26mg/6\n",
      "1.26mg/6\n",
      "0.63 mg/3\n",
      "1.2/1\n"
     ]
    }
   ],
   "source": [
    "meds['VALUE'] = meds['DOSE_VAL_RX'].map(find_mean_dose)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "meds['FEATURE_NAME'] = \"MED_\" + meds['GSN'].astype(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "dropper = [col for col in meds.columns if col not in {'SUBJECT_ID', 'DATE', 'FEATURE_NAME', 'VALUE'}]\n",
    "meds_preprocessed = meds.drop(columns=dropper).rename(columns=renamer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>DATE</th>\n",
       "      <th>VALUE</th>\n",
       "      <th>FEATURE_NAME</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>6</td>\n",
       "      <td>2175-06-12</td>\n",
       "      <td>2.0</td>\n",
       "      <td>MED_021796</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6</td>\n",
       "      <td>2175-06-12</td>\n",
       "      <td>5.0</td>\n",
       "      <td>MED_006562</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>6</td>\n",
       "      <td>2175-06-12</td>\n",
       "      <td>25000.0</td>\n",
       "      <td>MED_006522</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>6</td>\n",
       "      <td>2175-06-12</td>\n",
       "      <td>250.0</td>\n",
       "      <td>MED_nan</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>6</td>\n",
       "      <td>2175-06-12</td>\n",
       "      <td>20.0</td>\n",
       "      <td>MED_008208</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   SUBJECT_ID        DATE    VALUE FEATURE_NAME\n",
       "0           6  2175-06-12      2.0   MED_021796\n",
       "1           6  2175-06-12      5.0   MED_006562\n",
       "2           6  2175-06-12  25000.0   MED_006522\n",
       "3           6  2175-06-12    250.0      MED_nan\n",
       "4           6  2175-06-12     20.0   MED_008208"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meds_preprocessed.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Notes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here we can preprocess notes. Later the same things can be done using Spark."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_notes(sample_ids: set = None,\n",
    "              note_path: str = \"NOTEEVENTS.csv.gz\",\n",
    "              chunksize: int = 10_000) -> pd.DataFrame:\n",
    "    \"\"\"Get all notes or only those relevant for the sample.\"\"\"\n",
    "    if sample_ids is None:\n",
    "        return pd.read_csv(os.path.join(PATH, note_path))\n",
    "    return get_data_for_sample(sample_ids, note_path, chunksize)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "47b114714f0141209f6dae792d503bf8",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "0it [00:00, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "notes = get_notes(SAMPLE_IDS)\n",
    "# notes.to_csv(os.path.join(PATH_PROCESSED, 'SAMPLE_NOTES.csv'), index=False)\n",
    "\n",
    "# notes = pd.read_csv(os.path.join(PATH_PROCESSED, 'SAMPLE_NOTES.csv'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ROW_ID</th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>HADM_ID</th>\n",
       "      <th>CHARTDATE</th>\n",
       "      <th>CHARTTIME</th>\n",
       "      <th>STORETIME</th>\n",
       "      <th>CATEGORY</th>\n",
       "      <th>DESCRIPTION</th>\n",
       "      <th>CGID</th>\n",
       "      <th>ISERROR</th>\n",
       "      <th>TEXT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>174</td>\n",
       "      <td>22532</td>\n",
       "      <td>167853.0</td>\n",
       "      <td>2151-08-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2151-7-16**]       Dischar...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>175</td>\n",
       "      <td>13702</td>\n",
       "      <td>107527.0</td>\n",
       "      <td>2118-06-14</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2118-6-2**]       Discharg...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>176</td>\n",
       "      <td>13702</td>\n",
       "      <td>167118.0</td>\n",
       "      <td>2119-05-25</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2119-5-4**]              D...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>177</td>\n",
       "      <td>13702</td>\n",
       "      <td>196489.0</td>\n",
       "      <td>2124-08-18</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2124-7-21**]              ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>178</td>\n",
       "      <td>26880</td>\n",
       "      <td>135453.0</td>\n",
       "      <td>2162-03-25</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2162-3-3**]              D...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ROW_ID  SUBJECT_ID   HADM_ID   CHARTDATE CHARTTIME STORETIME  \\\n",
       "0     174       22532  167853.0  2151-08-04       NaN       NaN   \n",
       "1     175       13702  107527.0  2118-06-14       NaN       NaN   \n",
       "2     176       13702  167118.0  2119-05-25       NaN       NaN   \n",
       "3     177       13702  196489.0  2124-08-18       NaN       NaN   \n",
       "4     178       26880  135453.0  2162-03-25       NaN       NaN   \n",
       "\n",
       "            CATEGORY DESCRIPTION  CGID  ISERROR  \\\n",
       "0  Discharge summary      Report   NaN      NaN   \n",
       "1  Discharge summary      Report   NaN      NaN   \n",
       "2  Discharge summary      Report   NaN      NaN   \n",
       "3  Discharge summary      Report   NaN      NaN   \n",
       "4  Discharge summary      Report   NaN      NaN   \n",
       "\n",
       "                                                TEXT  \n",
       "0  Admission Date:  [**2151-7-16**]       Dischar...  \n",
       "1  Admission Date:  [**2118-6-2**]       Discharg...  \n",
       "2  Admission Date:  [**2119-5-4**]              D...  \n",
       "3  Admission Date:  [**2124-7-21**]              ...  \n",
       "4  Admission Date:  [**2162-3-3**]              D...  "
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "notes.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "notes['DATE'] = pd.to_datetime(notes['CHARTDATE']).dt.date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "def clean_text(note: str):\n",
    "    cleaned = re.sub(r'[^\\w]', ' ', note).replace(\"_\", \" \")\n",
    "    removed_spaces = re.sub(' +', ' ', cleaned)\n",
    "    lower = removed_spaces.lower()\n",
    "    return lower"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "notes['CLEAN_TEXT'] = notes['TEXT'].map(clean_text)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ROW_ID</th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>HADM_ID</th>\n",
       "      <th>CHARTDATE</th>\n",
       "      <th>CHARTTIME</th>\n",
       "      <th>STORETIME</th>\n",
       "      <th>CATEGORY</th>\n",
       "      <th>DESCRIPTION</th>\n",
       "      <th>CGID</th>\n",
       "      <th>ISERROR</th>\n",
       "      <th>TEXT</th>\n",
       "      <th>DATE</th>\n",
       "      <th>CLEAN_TEXT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>174</td>\n",
       "      <td>22532</td>\n",
       "      <td>167853.0</td>\n",
       "      <td>2151-08-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2151-7-16**]       Dischar...</td>\n",
       "      <td>2151-08-04</td>\n",
       "      <td>admission date 2151 7 16 discharge date 2151 8...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>175</td>\n",
       "      <td>13702</td>\n",
       "      <td>107527.0</td>\n",
       "      <td>2118-06-14</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2118-6-2**]       Discharg...</td>\n",
       "      <td>2118-06-14</td>\n",
       "      <td>admission date 2118 6 2 discharge date 2118 6 ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>176</td>\n",
       "      <td>13702</td>\n",
       "      <td>167118.0</td>\n",
       "      <td>2119-05-25</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2119-5-4**]              D...</td>\n",
       "      <td>2119-05-25</td>\n",
       "      <td>admission date 2119 5 4 discharge date 2119 5 ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>177</td>\n",
       "      <td>13702</td>\n",
       "      <td>196489.0</td>\n",
       "      <td>2124-08-18</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2124-7-21**]              ...</td>\n",
       "      <td>2124-08-18</td>\n",
       "      <td>admission date 2124 7 21 discharge date 2124 8...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>178</td>\n",
       "      <td>26880</td>\n",
       "      <td>135453.0</td>\n",
       "      <td>2162-03-25</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Discharge summary</td>\n",
       "      <td>Report</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>Admission Date:  [**2162-3-3**]              D...</td>\n",
       "      <td>2162-03-25</td>\n",
       "      <td>admission date 2162 3 3 discharge date 2162 3 ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ROW_ID  SUBJECT_ID   HADM_ID   CHARTDATE CHARTTIME STORETIME  \\\n",
       "0     174       22532  167853.0  2151-08-04       NaN       NaN   \n",
       "1     175       13702  107527.0  2118-06-14       NaN       NaN   \n",
       "2     176       13702  167118.0  2119-05-25       NaN       NaN   \n",
       "3     177       13702  196489.0  2124-08-18       NaN       NaN   \n",
       "4     178       26880  135453.0  2162-03-25       NaN       NaN   \n",
       "\n",
       "            CATEGORY DESCRIPTION  CGID  ISERROR  \\\n",
       "0  Discharge summary      Report   NaN      NaN   \n",
       "1  Discharge summary      Report   NaN      NaN   \n",
       "2  Discharge summary      Report   NaN      NaN   \n",
       "3  Discharge summary      Report   NaN      NaN   \n",
       "4  Discharge summary      Report   NaN      NaN   \n",
       "\n",
       "                                                TEXT        DATE  \\\n",
       "0  Admission Date:  [**2151-7-16**]       Dischar...  2151-08-04   \n",
       "1  Admission Date:  [**2118-6-2**]       Discharg...  2118-06-14   \n",
       "2  Admission Date:  [**2119-5-4**]              D...  2119-05-25   \n",
       "3  Admission Date:  [**2124-7-21**]              ...  2124-08-18   \n",
       "4  Admission Date:  [**2162-3-3**]              D...  2162-03-25   \n",
       "\n",
       "                                          CLEAN_TEXT  \n",
       "0  admission date 2151 7 16 discharge date 2151 8...  \n",
       "1  admission date 2118 6 2 discharge date 2118 6 ...  \n",
       "2  admission date 2119 5 4 discharge date 2119 5 ...  \n",
       "3  admission date 2124 7 21 discharge date 2124 8...  \n",
       "4  admission date 2162 3 3 discharge date 2162 3 ...  "
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "notes.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Notes processing\n",
    "Let's see if we can choose some subparts of notes that will contain more information."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Nursing/other        822497\n",
       "Radiology            522279\n",
       "Nursing              223556\n",
       "ECG                  209051\n",
       "Physician            141624\n",
       "Discharge summary     59652\n",
       "Echo                  45794\n",
       "Respiratory           31739\n",
       "Nutrition              9418\n",
       "General                8301\n",
       "Rehab Services         5431\n",
       "Social Work            2670\n",
       "Case Management         967\n",
       "Pharmacy                103\n",
       "Consult                  98\n",
       "Name: CATEGORY, dtype: int64"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "notes.CATEGORY.value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see how how often patients have different kinds of notes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "has_note_type = notes.groupby(['SUBJECT_ID', 'CATEGORY']).HADM_ID.count().unstack().fillna(0) > 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "CATEGORY\n",
       "Discharge summary    0.891237\n",
       "Radiology            0.789841\n",
       "ECG                  0.740238\n",
       "Nursing/other        0.648052\n",
       "Echo                 0.441988\n",
       "Nursing              0.165518\n",
       "Physician            0.163958\n",
       "Respiratory          0.077840\n",
       "General              0.062389\n",
       "Nutrition            0.061024\n",
       "Rehab Services       0.045919\n",
       "Social Work          0.028757\n",
       "Case Management      0.012330\n",
       "Pharmacy             0.001452\n",
       "Consult              0.001062\n",
       "dtype: float64"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(has_note_type.sum() / has_note_type.shape[0]).sort_values(ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see if having some kind of notes correlates with mortality"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "has_note_type['DECEASED'] = has_note_type.index.isin(DECEASED_TO_DATE)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7fed104ff310>"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAApEAAAHSCAYAAAC0D5CUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAgAElEQVR4nOzde3zP9f//8dt7R1szW8vIJtX0Qc4dbI0PWT6RbcbIeSStExafWs5aIaKIkU8kSRQyGzmsg5RGcyit5FA5fK3YiJlhbHu/f3/4eWdtxvbe+2XN/Xq5vC6X9+v4eL7eS3vseTRZLBYLIiIiIiKl4HC9CyAiIiIi/zxKIkVERESk1JREioiIiEipKYkUERERkVJTEikiIiIipaYkUkRERERKTUmkiIiISCWSk5NDWFgY6enpRc7t3r2byMhI2rdvz+jRo8nPzy9zHCWRIiIiIpXEDz/8QK9evTh48GCx52NjYxk3bhzJyclYLBaWLVtW5lhKIkVEREQqiWXLlvHSSy/h6+tb5Nzvv/9Obm4uzZo1AyAyMpL169eXOZZTme8UEREREbvLzs4mOzu7yHFPT088PT0LHZs4ceIVn5OZmUn16tWt+9WrVycjI6PM5VISKeQd329YrJn3jDMsFkCBydBwOFfyRUTbWE4bGi+vwNHQeO+6GBfvGct5w2IB7M6rami8OzlnaLwNzu6GxmueW/Z+ZGVxk8nYeBcsxv7b+8nV2dB4z/3fB4bGs/X37MIP1zBr1qwixwcPHsyQIUOu+TlmsxmT6a9fjBaLpdB+aSmJFBEREanA+vfvT5cuXYoc/3st5NXUrFmTY8eOWfePHz9ebLP3tVISKSIiImJP5gKbbi+u2bos/Pz8cHV1ZceOHdx7770kJSXRunXrMj9PA2tERERE7Mlitm2zUXR0ND/++CMAr7/+OpMmTaJDhw6cPXuWfv36lfm5qokUERERsSez7YlgaW3YsMH6ed68edbP9evX5+OPPy6XGKqJFBEREZFSs3sSmZOTw8svv0xYWBgRERFERUWxa9cuu8WLj4+nXr16fP/994WOT5w4kXr16tkt7vV0+vRpBg0adL2LISIiIsWwWMw2bRWVXZNIs9lMdHQ01apVIzExkaSkJAYNGkR0dDQnT560W9yaNWuSnJxs3bdYLGzbts1u8a63U6dOsXv37utdDBERESmO2WzbVkHZtU9kamoqR44cISYmBgeHi/lqUFAQkyZNwmw2k5+fT1xcHL/88gvHjx+nXr16TJs2jfz8fP773/9y/PhxAAYNGsRDDz3EoUOHiIuLIysriypVqjB27FjuvvvuInEfeughvvjiC0aMGAHA9u3badasmTXRysnJYdSoUWRkZJCZmckDDzzAxIkT2bp1K2+//TZVqlTht99+o169erz++uu4uLgwffp0tmzZwqlTp/D19WX69OnccsstrF27lpkzZ+Lu7k6DBg0oKChg8uTJpKWlMWnSJHJzc/H29ubll1+mdu3aREVFcffdd7Njxw7Onz/PCy+8wPvvv89vv/3GY489xmOPPcaZM2d45ZVX+OWXXygoKCA6OpqwsDASEhLYtGkTp06d4vDhw7Rs2ZK4uDgmTJhAZmYmgwYNYvbs2fb8kYqIiEhpVeDaRFvYtSby559/pn79+tYE8pI2bdrg4+PD999/j7OzM0uXLuWzzz7j9OnTfPXVV3z22Wf4+fmRkJDAxIkT2b59OwDDhw8nNjaWlStXMn78eIYNG1ZsXG9vb2rXrk1aWhoAa9eupWPHjtbzGzdupEGDBixdupTk5GS2bdtmbWL//vvvGTduHOvWreOPP/7gm2++4dChQ+zfv5+PPvqI5ORkbr31VlatWsWJEyd49dVXWbhwIR9//DGnTp0C4MKFC4wZM4Y33niDlStXMmDAAMaOHWuNb7FY+Pjjj2nfvj0TJkxg1qxZLF682JoAzpkzh4YNG5KQkMDixYv53//+x+HDh63lmzlzJqtWreLLL79k7969jBkzBl9fXyWQIiIiYhi71kQ6ODjg6up6xfP3338/Xl5eLF68mP3793Pw4EHOnj1L8+bNmTZtGhkZGTz44IMMGjSIM2fO8NNPPzFy5Ejr/WfPnuXkyZN4e3sXefYjjzxCcnIyDRs25Pvvvy+UxIWFhZGWlsZ7773H/v37ycrK4uzZswDcdddd1KxZE4CAgABOnTpFnTp1GD58OMuXL+fAgQPs3LmT2267je3bt9O8eXNq1KgBQOfOnfn88885ePAghw8f5plnnrHGzMnJsX6+NCdTrVq1aNq0KW5ubvj5+VmXNNq8eTO5ubmsWLHC+p6//PILAM2bN8fDwwOA2rVrc+rUKW666aZr/ZGIiIiI0WycJ7KismsS2ahRI5YsWVJkWZ1p06YRHBzMmTNnmDlzJv369SMyMpKTJ09isVi4/fbbWbduHZs2beLLL7/k3XffZfny5bi4uJCUlGR9ztGjR/Hy8io2drt27ejVqxetWrXivvvuK1QbumjRIpKTk+nevTvBwcHs27cPi+XienWXJ70mkwmLxcJPP/3E888/z2OPPUb79u1xcHDAYrHg4OCAuZi+CmazGX9/f2tZCwoKrE3zAM7Ofy3v5ORU9EdgNpuZOnUqDRs2BC7OKF+tWjVWr15dbPlERESkAlNzdundd999+Pj4MGvWLAoKLmbhmzZtIiEhgbp167JlyxYeeeQRunbtiqenJ6mpqRQUFPDBBx8QHx/PI488wksvvcSJEyesyeWlxCwlJYU+ffpcMba3tzd+fn7MmDGjUFP2pXt79OhBp06dOH/+PHv27Ck2Gbxk27ZttGjRgl69enH77bezceNGCgoKuOeee/jxxx/JzMzEYrGwdu1aTCYTd955J6dOnbI2w69YsYIXXnjhmr+3oKAgPvzwQ+DiYumdOnXiyJEjV7zeycmJ/Hxj11UVERGRa6SBNaVnMpl46623mDRpEmFhYTg5OeHt7c3cuXO55ZZbePTRR3nhhRdYs2YNzs7O3HPPPaSnpxMdHc1///tfwsPDcXR0JDY2Fk9PT6ZOnUpcXBzvvPMOzs7OTJ8+vcSFwzt06MDs2bNp3rx5oeP9+/cnLi6OuXPn4uHhQfPmzUlPT+e2224r9jkdO3Zk8ODBhIeHAxdrWNPT07n55psZM2YMjz/+OC4uLvj7++Pp6YmLiwszZsxg4sSJnD9/Hg8PD1577bVr/t4GDx5MXFwcYWFhFBQUEBsba20+L46Pjw+1atUiKiqKRYsWXXMcERERsb+KPE2PLUwWtYeW2cmTJ1m0aBGDBw/GwcGBCRMmUKdOHaKioq530Uol7/h+w2LNvGecYbEACq78N4ZdOFfyf01tLKcNjZdX4GhovHddjIv3jOW8YbEAdudVNTTenZwzNN4GZ3dD4zXPNbb15yaTsfEuWIz9t/eTq/PVLypHz/3fB4bGO//btzbd7xoQVE4lKV9a9tAGXl5eZGdnExYWhqOjIw0bNqR79+7Xu1giIiJSkVTgJmlbKIm0gclkYsyYMde7GCIiIlKRVdLmbCWRIiIiIvakKX5EREREpNQqaU2kXaf4EREREZHKSTWRYuiI6ZjvXjEsFsDke8de/aJylGf0aHCMDbjDbOwI3zxjB4jihXFNTqkFV17Nyx5OGTv4lT/Nxo6WzsPYqRF+dTH216eTwb+u8wyNBjmmSj61hQbWiIiIiEipVdLmbCWRIiIiIvZUSWsi1SdSREREREpNNZEiIiIidmSxaIofERERESkt9YkUERERkVJTn0i5mpycHF5++WXCwsKIiIggKiqKXbt22T1ufHw88fHxAIwcOZLff//d7jFFRETkGlnMtm0VlJLIcmI2m4mOjqZatWokJiaSlJTEoEGDiI6O5uTJk4aVIzU1FYulks+3JSIiItedkshykpqaypEjR4iJicHJ6WIvgaCgICZNmoTZbOZ///sfHTt2JDw8nMmTJ1NQUEB6ejqdO3cmNjaWsLAw+vfvT1ZWFnl5ecTGxtK5c2c6d+7MsmXLABgxYgQJCQnWmPXq1StUhrlz55KZmcmTTz5paOIqIiIiJTAX2LZVUEoiy8nPP/9M/fr1cXAo/JW2adOGn376iQ0bNrBixQpWrlzJoUOH+OijjwDYs2cPAwYM4JNPPsHT05PVq1fz/fffc+rUKRITE3n77bfZvn37NZXhySefxNfXl7lz5+Lt7V3u7ygiIiJloOZsKYmDgwOursUvY/btt98SGhqKm5sbTk5OdO3alS1btgDg4+PD3XffDcBdd93FqVOnuOuuuzhw4AADBw5k/fr1vPjii4a9h4iIiJQzs9m2rYJSEllOGjVqxM8//1ykP+K0adOsCePl8vPzAQolniaTCYvFgre3N2vWrKFv374cOHCALl26kJ2dbT0PkJdn9MqmIiIiUiaqiZSS3Hffffj4+DBr1iwKCi72X9i0aRMJCQn079+fNWvWkJubS35+PitWrCAoKOiKz/riiy+IjY3lwQcfZMyYMbi7u3PkyBG8vLz49ddfAfj888+LvdfR0dEaX0RERMReNE9kOTGZTLz11ltMmjSJsLAwnJyc8Pb2Zu7cudx9990cOXKErl27kp+fT6tWrejbty9Hjx4t9lmtW7fm008/JTQ0FFdXVzp16kS9evXo1asXQ4cOJTw8nKCgIKpXr17k3gcffJAnn3ySd955h9q1a9v7tUVERORqKnCTtC1MFs0Hc8N747a+hsWK+e4Vw2IBTL53rKHxjK7ad8ZkaDxvgyu584x9PQ45GveCd+Y7GhYL4JSx4XA3+HfmCQdjf5XdbDb2P06ja3yM7jCVbfDPb/ShxYbGy920yKb7q/w7qpxKUr5UEykiIiJiR5V17Wz1iRQRERGRUlNNpIiIiIg9VdI+kUoiRUREROypAk/TYwslkUKBgf3DjR7oMmLHeEPj5Tz1uKHxDm6rZmi8VFNVQ+N5GtyNaAcnDIt1h6no7Ar2dB5jBy7UMvJ/LEDPRumGxvvoJ2NnvzAZPAT2gsGd3e7NrZx9Bq0MrIlcvXo1c+bMIT8/n/79+9OnT59C53ft2sW4cePIy8vj1ltvZerUqXh6epYplvpEioiIiNiTQZONZ2RkMH36dJYsWUJiYiJLly61zi99ycSJE4mJiWHVqlXccccdzJ8/v8yvpSRSREREpBLYvHkzQUFBeHl54e7uTvv27Vm/fn2ha8xmM2fOnAHg3LlzVKlSpczx1JwtIiIiYk82NmdnZ2eTnZ1d5Linp2ehpujMzMxCC5H4+vqSlpZW6J4RI0bw+OOP8+qrr+Lm5sayZcvKXC4lkSIiIiL2ZOPAmoULFzJr1qwixwcPHsyQIUOs+2azGZPpr/7IFoul0H5ubi6jR4/mvffeo0mTJixYsIDhw4czd+7cMpVLSaSIiIiIPdlYE9m//2N06dKlyPG/D4ipWbMm27dvt+4fO3YMX19f6/6+fftwdXWlSZMmAPTo0YMZM2aUuVzqEykiIiJSgXl6euLv719k+3sSGRwczJYtWzhx4gTnzp3j008/pXXr1tbzderU4ejRo+zfvx+AL774gsaNG5e5XHapiUxPT6dDhw4EBAQAF6tP77nnHp5//nluueUWfvzxRz766CMmTpxYqueGhITw/vvv4+/vb49ii4iIiJQ/g6b4qVGjBsOGDaNfv37k5eXRrVs3mjRpQnR0NDExMTRu3JhJkyYxdOhQLBYLPj4+vPrqq2WOZ7fmbF9fX5KSkoCLbfLTpk0jJiaGJUuW0LhxY5syXxEREZF/DAMnGw8PDyc8PLzQsXnz5lk/t2nThjZt2pRLLEP6RJpMJoYMGULLli3Zs2cPp06dYtasWSxatIgFCxawcuVKHBwcaNKkCa+88grnz5/n5ZdfZseOHTg7O/Pss8/SsWNHAGbPns3u3bs5d+4cU6ZMoWnTpmzdupXp06eTm5tLdnY2I0eOpF27dowYMYKsrCwOHTpEbGwsN910ExMmTMDR0ZFmzZrx22+/sWjRIg4dOkRcXBxZWVlUqVKFsWPHcvfddxd6h9WrV/POO+/g6OiIv78/U6dOZefOndb3gIsjnlq0aEGLFi0YNGgQd955J7/++it33303zZs3Z+XKlZw6dYrZs2cTEBBASEgIoaGhpKSk4OTkxLPPPsu7777LoUOHGD58OB07dmTfvn2MHz+es2fPcuLECZ588kl69epFfHw8O3fu5MiRI/Tq1Yv58+ezYcMGHBwcSE1NZd68ebzzzjtG/HhFRESkJJV02UPD+kS6uLhQp04dazs8QEFBAW+//TYrVqwgISGBvLw8MjIyWLRoEWfPnmXdunUsWLCA2bNnc+HCBQDq1q1LYmIiUVFR1gkyP/jgAyZMmMDKlSuZMGFCoU6iXl5erFu3jn//+9+8+OKLTJ06lcTERJyc/sqfhw8fTmxsLCtXrmT8+PEMGzasSPnffPNN3n33XRISEvDz8yv0HsXZu3cv0dHRJCUl8d133/H777+zdOlSwsLCWLp0qfW6W265hYSEBAICApg7dy7vvvsuU6dOtY6UWr58Oc8++ywrVqzg/fffZ8qUKdZ7L1y4wNq1a4mKisLf35/U1FQAEhMTiYyMvOafjYiIiNiRQZONG83Q0dkmk6nQpJaOjo40b96cbt268dBDDzFgwABq1KjBtm3b6N69Ow4ODlSvXp01a9ZY72nXrh1wMZlMTk4GYOrUqXz55ZesX7+eH374wTqJJmAdgbRv3z58fHyoX78+AN26dWPixImcOXOGn376iZEjR1rvOXv2LCdPnsTb29t6rG3btvTq1Yt27drRvn17GjRoYE3ainPLLbdYazNr1qzJAw88AECtWrVIT/9rea5LHV5r1aqFr68vTk5O1KpVyzof1IgRI9i0aRNvv/02+/bt4+zZs0XeDaBr166sWrWKZs2a8e233xIXF3flH4SIiIiIjQyribxw4QIHDhygbt26hY6/9dZbxMXFYbFYeOKJJ9i6dStOTk6F5jU6dOiQtSbS0dERoND53r17k5aWRqNGjXj66acLPf9S0uro6Ii5mOpks9mMi4sLSUlJ1m358uV4eXkVum7MmDHMnDmTatWqERsbS1JSEiaTCYvlrwVN8/LyrJ9dXFwK3X+p3H/n7Oxs/Xx57eglQ4cO5bPPPiMgIIChQ4cW+24AHTp0ICUlheTkZFq3bo2rq2ux8URERMRgZrNtWwVlSBJpNpuJj4+nadOm3HbbbdbjJ06coGPHjvzrX//iueeeo2XLluzdu5f777+ftWvXYrFY+PPPP+nbt681ify7rKwsDh48yHPPPUfr1q354osvKCgoupD7nXfeSXZ2Nnv37gUu9nEEqFq1Krfffrt1EFBKSkqRxcrz8/N5+OGH8fb25qmnniIiIoLdu3fj7e3N4cOHOX/+PFlZWezYsaNcvq/LpaSkEBMTQ7t27fj6668Bin0/Nzc3WrduzbRp09SULSIiUpGoObt0MjMziYiIAC4mkQ0aNGDatGmFrrn55pvp0aMH3bp1w83NjTvuuIOuXbvi5OTEhAkT6NSpEwBjx47Fw8Oj2DheXl5069aN0NBQnJycCAoKIjc3t1CzL1ysGZwyZQrDhw/HwcGBO+64w1qTN3XqVOLi4njnnXdwdnZm+vTphWo6nZyciImJ4fHHH8fV1RUfHx8mT56Mj48Pbdq0ITQ0FD8/P+69995y+/4uGTJkCL1798bV1ZX69evj5+dXqDn8cqGhoXz33Xc0bdq03MshIiIiZVSBaxNtYbJc3h5biZnNZl5//XUGDx6Mu7s7CxYsICMjgxEjRlzvopWLgoICpk+fjo+PDwMGDCjVvVPq9LVTqYo6j7H/uY3YMd7QeDlPPW5ovIPbqhkaL9VS1dB4HkUr3e3qPcdjhsXqRvWrX1SOjjkY+2/vjjzT1S8qRx2aHjY03kc/1TY0nrPBv6lzDF6KpOH5fEPjdcj4yNB455a9YtP9bt3HlVNJytcNs+yhg4ODtdbS2dkZPz+/Uk92XpF17doVb29v5syZc72LIiIiIjeAGyaJBHjyySd58sknr3cx7CIxMfF6F0FERESKU0kbfW+oJFJERETEcJW0T6SSSBERERF7qqRJpMFdZ0VERESkMlBNpBg66s/gAZuGj5b2ePtdQ+P59yrdSHxb7dpl8Ohsg/96L3AwLp67wV2kHAyuMjD6Z3f8t5sMjedm8M/P1eCKrAsG/7+6hus5YwMarQLP9WgLJZEiIiIi9lRJm7OVRIqIiIjYk0Zni4iIiEipVdKaSA2sEREREZFSU02kiIiIiD1V0ppIJZEiIiIi9qTR2WIP6enpdOjQgYCAgELHu3fvTu/evXnvvfesSxo6ODjwxBNPEBoaar1u3bp1zJ8/nzNnzpCXl0eLFi0YOXIkVasaOxWLiIiIFM9i1sAasRNfX1+SkpKKHJ82bRo///wzH3zwAVWrVuXo0aP07dsXb29vgoODWb16NbNmzeKtt94iICAAi8XC1KlTGT16NDNnzrwObyIiIiJFqDlbjHTmzBkWLlzIqlWrrLWKNWvWZNq0abi5uQEwa9YsRo0aZa3FNJlMDBs2jPfee+96FVtERERuEEoiK4DMzEwiIiIKHRs7dixOTk7UqVOn0PEmTZoAkJWVxcGDB7nvvvsKnXd2diY6Otq+BRYREZFrpz6RYi/FNWfv2rULFxeXq95rMl1cmyo9PZ1BgwYBcOLECZYtW8att95a/oUVERGR0qmkfSI1T2QFFRAQQG5uLn/88Ueh42vWrGHhwoV4eXlRu3ZtvvvuOwD8/f1JSkoiKSkJZ2dnCgoKrkexRURE5O/MZtu2CkpJZAVVpUoV+vTpQ1xcHDk5OcDF2sZp06ZZ+0AOHTqUCRMm8Ntvv1nv2759O1lZWTg6Ol6XcouIiMiNQc3ZFUBxfSLvv/9+Ro4cyezZs+nevTtOTk44Ojry/PPP06pVKwDCwsJwd3dnzJgxnDlzhpycHAICApg1a5aaskVERCqKClybaAslkdeZv78/P/300xXPx8TEEBMTc8XzISEhhISE2KNoIiIiUh4slbNPpJJIEREREXtSTaSIiIiIlJpGZ4uIiIiIXKSaSBERERF70mTjIrZzxmRovIPbqhkaz7/XAEPjeX24wNB4NBlnaLiTjsY2ltzm5GlYLHOeYaEAcDT4396fTgb/Wz9l3M8OwNHg357GfpuQb3DA/7twk6HxmhsajUrbnK0kUkRERMSOLBpYIyIiIiKlVklrIjWwRkRERERKTUmkiIiIiD1ZzLZtpbB69Wo6duzIww8/zOLFi4uc379/P1FRUXTq1ImBAwdy6tSpMr+WkkgRERERezJbbNuuUUZGBtOnT2fJkiUkJiaydOlSfv31V+t5i8XCM888Q3R0NKtWraJBgwbMnTu3zK+lJFJERETEnsxm27ZrtHnzZoKCgvDy8sLd3Z327duzfv166/ldu3bh7u5O69atAXj66afp06dPmV9LA2uuo/T0dDp06EBAQECh4927dy/2hxofHw/AkCFDDCmfiIiIXH/Z2dlkZ2cXOe7p6Ymn51/TW2VmZlK9enXrvq+vL2lpadb9//u//+OWW25h1KhR7N69mzvvvJOxY8eWuVxKIq8zX19fkpKSrncxRERExF5sHJ29cOFCZs2aVeT44MGDC1Usmc1mTKa/Jvm0WCyF9vPz89m6dSsffPABjRs35s0332Ty5MlMnjy5TOVSEllBrV69mjlz5mAymWjcuDHjx48HIC0tjZ49e5KRkUFkZCRDhgzBbDbz6quvsmXLFkwmE506deLJJ5+8zm8gIiIigM0r1vTv358uXboUOX55LSRAzZo12b59u3X/2LFj+Pr6WverV69OnTp1aNy4MQBhYWHExMSUuVxKIq+zzMxMIiIiCh0bPXo0kyZNIiEhgZo1axIbG8tXX30FwJ9//slHH31ETk4OISEhDBgwgKSkJI4cOcKqVau4cOECUVFR/Otf/+LBBx+8Dm8kIiIihdhYE/n3ZusrCQ4OJj4+nhMnTuDm5sann35qrYQCaN68OSdOnGDPnj3Ur1+fDRs20LBhwzKXS0nkdVZcc/b69eu55557qFmzJgBTp04FYPfu3fz73//GxcWFm2++GW9vb06dOkVqaipdunTB0dERNzc3wsPD2bJli5JIERGRCsCoFWtq1KjBsGHD6NevH3l5eXTr1o0mTZoQHR1NTEwMjRs3Zvbs2YwZM4Zz585Rs2ZNpkyZUuZ4SiIrICcnp0J9GE6cOFHo3CUmkwmLxYL5b/9xWiwWCgoK7F9QERERqVDCw8MJDw8vdGzevHnWz02bNuXjjz8ul1ia4qcCaty4MTt37uTYsWMAvPrqq3zxxRdXvD4oKIjExEQKCgo4d+4cq1evJjAw0KjiioiISEkMmifSaKqJvM6K6xN5//33M3r0aAYOHIjZbKZZs2ZERkby1ltvFfuMHj16cPDgQSIiIsjLyyM8PJz//Oc/RhRfRERErqYCJ4K2UBJ5Hfn7+/PTTz9d8XyHDh0K7f99fsgNGzZYP48ZM6Z8CyciIiLlw8bR2RWVkkgRERERe6qkNZHqEykiIiIipaaaSBERERE7slTSmkglkSIiIiL2pCRSRERERErNoMnGjaYkUmhjOW1YrB3mqobFAkg1GRtv1y5j49FknKHheqS9Ymi881NfMDTe24tOXP2icjIr0t2wWACrEm42NF5E1FlD483/0Njvs8ft6YbGc2/gami8M7suGBpv1UE/Q+NJ+VASKSIiImJPas4WERERkVJTEikiIiIipWWxKIkUERERkdKqpDWRmmxcREREREpNNZHlID8/n3nz5rFq1SpMJhMFBQV06dKFp556CpPJZGhZEhIS2Lp1K5MnTzY0roiIiFxBJa2JVBJZDl5++WWOHz/O0qVL8fT0JCcnh0GDBlG1alX69OlzvYsnIiIi15FWrJFiHT16lFWrVvH111/j6ekJgIeHB+PGjePXX3/l+PHjjBs3jqNHj2IymXj++ecJDg4mPj6ejIwMDh06xO+//86jjz7KM888Q0FBAVOmTGHr1q0UFBQQGRnJY489RmpqKlOnTsVsNnPXXXfx3//+l1GjRnH69GkyMzPp0qULzz333HX+Ns5gco0AACAASURBVERERKQIJZFSnLS0NAICAqhWrVqh4wEBAQQEBDBs2DC6du3KQw89RGZmJr179yYxMRGAvXv3snjxYk6fPk27du3o06cPa9asAWDlypVcuHCBgQMH0qhRIwAOHjzIl19+SdWqVZk/fz5hYWF06dKF06dP06ZNG6Kioox9eREREbm6yrlgjZLI8nB5v8f169czZ84czGYzLi4upKens3//fmbOnAlc7D95+PBhAAIDA3FxccHHxwcvLy9Onz7Nli1b2L17N99++y0AZ8+eZe/evdStW5c77riDqlUvrogycOBAvv32W+bPn88vv/xCXl4e586dM/jNRURE5EalJNJGjRo14rfffiMnJwcPDw86dOhAhw4dSE9Pp1+/fpjNZhYuXIiXlxcAmZmZ+Pj48Pnnn+Pq+tcyViaTCYvFQkFBAbGxsTz88MMAnDhxgptuuomdO3dSpUoV6/WTJ0/m8OHDhIWF0a5dOzZv3lxp56ESERH5J6usfSI1xY+NatWqRadOnRg+fDjZ2dnAxdrGjRs34uDgQFBQEEuWLAHg119/JTw8vMQaw6CgIJYtW0ZeXh5nzpyhd+/e7Ny5s8h1KSkpDBw4kEceeYQDBw6QkZGBuZIu8C4iIvKPZrbYtlVQqoksB3FxcSxYsIB+/fpRUFDAmTNnCAwMZN68ebi7uzNu3DjCw8MBmDJlCh4eHld8Vs+ePTl06BBdunQhPz+fyMhIAgMDSU1NLXTdU089xYsvvkiVKlWoWbMmjRo1Ij093a7vKSIiImVQSet4TBa1gd7wvqsdYVisHZaqhsUCKDB2mk48CoyNZ7Qeaa8YGu/81BcMjddu0QnDYn0W6W5YLIBVCTcbGi8i6qyh8eZ/aOz32eN2Y/9od2/gevWLytGZXRcMjbfqoJ+h8Z5K/8DQeFk92tp0v9fSL8upJOVLNZEiIiIidlRZ+0QqiRQRERGxp0ranK0kUkRERMSOVBMpIiIiIqVXSWsiNcWPiIiIiJSaaiKFvAJH42IZFwoAT4NHS3sYPFfnSUdj/w40erS0a+zrhsbbPv3fhsUyubUwLBZAjsFVBq7Pv2ZovCqLXzY2Xm1jv1CnercZGs+c9puh8U4Z/LvBaJZKWhOpJFJERETEnpREioiIiEhpqSZSREREREqvkiaRGlgjIiIiIqWmmkgRERERO1JztoiIiIiUWmVNItWcXU7S09OpV68eKSkphY6HhISQnp5u07MjIiJsul9ERESuH4vZtq00Vq9eTceOHXn44YdZvHjxFa/buHEjISEhNr2Xkshy5OzszNixY8nJySnX5yYlJZXr80RERKTyycjIYPr06SxZsoTExESWLl3Kr7/+WuS648eP89prts/lWmISee7cOZsD3Eh8fX0JDg4u8oPZunUrUVFR1v0RI0aQkJBAeno6HTp0oFevXgwYMIA9e/bQvXt3IiMj6dWrFwcPHgSgXr16AMTHxzNmzBiioqIICQlhzpw5AOTl5TFq1Cjat29Pv3796N+/P6mpqca8tIiIiJTMYrJtu0abN28mKCgILy8v3N3dad++PevXry9y3ZgxYxg8eLDNr1Vin8iuXbsyY8YM7rrrLpsD3ShGjBhBeHg4KSkptGzZ8qrXHzhwgHfeeQd/f39GjhzJgAEDeOSRR1i5ciU7d+7k9ttvL3T93r17Wbx4MadPn6Zdu3b06dOHpKQkzp07x/r16/njjz8IDw+309uJiIhIadnaJzI7O5vs7Owixz09PfH09LTuZ2ZmUr16deu+r68vaWlphe55//33ufvuu2natKltheIqNZF9+vQhKiqK5cuX2xzoRuHh4cH48eOvuVnbx8cHf39/ANq0acP48eMZNWoUVatWLTYZDAwMxMXFBR8fH7y8vDh9+jQpKSmEh4djMpnw8/PjgQceKPf3EhERkbKxmE02bQsXLuShhx4qsi1cuLBQHLPZjMn0V82lxWIptL9v3z4+/fRTnn322XJ5rxJrIvv06UPLli0ZPnw43377LRMmTMDNza1cAldmrVq1KtSs7e/vj8VisZ7Py8uzfq5SpYr1c4cOHWjevDlffvkl7733Hhs3bmTChAmFnu3q6mr9bDKZsFgsODo6YjZ4zWYRERG5NrbWRPbv358uXboUOX55LSRAzZo12b59u3X/2LFj+Pr6WvfXr1/PsWPH6Nq1K3l5eWRmZtK7d2+WLFlSpnJddWDN7bffzocffoizszP/+c9/CA8Pt25yZSNGjOCbb74hMzMTb29vDh8+zPnz58nKymLHjh3F3jN06FB+/PFHevbsyXPPPcfPP/98TbGCg4NZu3YtFouFjIwMtm7dWugvDxEREfnn8vT0xN/fv8j29yQyODiYLVu2cOLECc6dO8enn35K69atredjYmJITk4mKSmJuXPn4uvrW+YEEq5hnsi8vDzefPNNvvrqK1544QVq165d5mA3kkvN2gMHDsTNzY02bdoQGhqKn58f9957b7H3PP3004wePZrZs2fj7OxMXFzcNcXq3r07e/bsITw8nOrVq1OrVq1CNZwiIiJy/VhKMTjGFjVq1GDYsGH069ePvLw8unXrRpMmTYiOjiYmJobGjRuXazyT5fJ21r/Zs2cPsbGxeHl5MXXqVGrWrFmuwaV8bNy4EYvFQtu2bTl9+jSdO3dmxYoVeHl5XdP9qbUi7VzCv+xwdDcsFoBngaHh8DC4W8FJR2Nn6Xq071lD47nGvm5oPLda/zYsVtZzLQyLBbD4w5sMjff4dy8ZGu/de142NF6PNn8YGs/l3gBD451c/puh8Rb/UcvQeC8e+sDQeOmBts3H6J+6oZxKUr5KrIns2bMnTz31FE8//bSaRyuwgIAAXnzxRd58803gYnX1tSaQIiIiYl8Wc+XMoUpMIhcuXHjFIeAXLlzAxcXFLoWS0qlduzYffvjh9S6GiIiI3EBKbAtr0KABy5YtIzk5udDxDRs2EBoaateCiYiIiFQGFottW0VVYk3kuHHj+OWXX8jJySE7O5v27dszfPhwNm/ezBNPPGFUGUVERET+sW7I5uzt27ezdu1aTp48ydChQ1mwYAG1atXik08+0SjtSuRdF0fDYnlh7EiXHZwwNF6Bg7EDa25z8rz6ReXo7UXGfp/bpxs30AXg3B+bDIvVsskAw2IBtHUxdsYGN/8HDY3X5db7DI33wUZjk4ITn6dd/aJyVMfVx9B4nSpwbVt5uCGTyJtuugkXFxdq1KjBwYMHiYqKKrdZzkVERERuBBW5SdoWJfaJvHxEtpeXlxJIEREREQGuUhN5eRKpyatFRERESu+GbM4+fPgwTz/9dJHPl/zvf/+zX8lEREREKgGjVqwxWolJ5OjRo62f27dvb/fCiIiIiFQ2FmPHXBqmxCSyS5cu1s+nT5/GZDLh4eFh90KJiIiIVBbmSloTedWFd7/++mvCwsJo0aIF999/P2FhYXz11VdGlM0m6enp1KtXj5SUlELHQ0JCSE9Pt+nZERERNt3fs2dPSliynKioKOvnevXq2RRLRERExB5KTCK3b99OXFwcQ4cOZfv27WzZsoVBgwbx0ksvsW3bNqPKWGbOzs6MHTuWnJyccn1uUlJSme89cOAAderUKXEt8q1bt5b5+SIiIlKxWCwmm7aKqsTm7Dlz5jB9+vRC62c/8sgj1KxZk/j4eO6//367F9AWvr6+BAcH89prrzF+/Hjr8a1btzJy5EgWLVoEwIgRI2jRogUtWrTgiSeewNvbmypVqjB8+HDGjRtHfn4+rq6uTJo0idtvv5169eqxd+9e4uPjycjI4NChQ/z+++88+uijPPPMM+Tl5fHSSy+xY8cOatSogclk4tlnnyUwMJCvv/6a1q1bAxcHJq1atQpHR0datmxJbGwskyZNAuDRRx9l+fLlwMWVg3bu3AlAfHw8derUIS0tjUmTJpGbm4u3tzcvv/wytWvXJioqimrVqvHLL7/w5ptv0qBBAyO/chEREfmbyjo6u8SayD///LNQAnlJ8+bNOXr0qN0KVZ5GjBjBN998U6RZ+0oOHDjA1KlTWbBgAQsXLmTAgAEkJCTQvXt3ayJ3ub179zJ//nyWL1/O3Llzyc7O5qOPPuLcuXOsX7+eSZMm8eOPP1qvT0lJoWXLlnz11Vds2LCBFStWsHLlSg4dOsRHH33EmDFjAKwJJEBwcDCrVq2iZcuWfPTRR1y4cIExY8bwxhtvsHLlSgYMGMDYsWOt19erV4/k5GQlkCIiIhXADbl29rlz5654zsHhqt0pKwQPDw/Gjx/P2LFjWbVq1VWv9/Hxwd/fH4A2bdrwyiuvsGnTJkJCQmjbtm2R6wMDA3FxccHHxwcvLy9Onz5NSkoK3bt3x2Qy4efnxwMPPABAbm4u58+fx8vLi2+//ZbQ0FDc3NwA6Nq1K4mJifTp06dIjHbt2gFQt25dtm/fzsGDBzl8+DDPPPOM9ZrLm+ybNGlSim9IREREpPRKTCJr1KhBWlpakaQkLS0NX19fuxasPLVq1crarA3g7+9faGBLXl6e9fPlk6p36NCB5s2b8+WXX/Lee++xceNGJkyYUOjZrq6u1s8mkwmLxYKjoyNmc9Hx/KmpqbRo0QKg2PP5+fnFlt/JyanQ881mM/7+/ta+mQUFBRw/frzYdxAREZHr64Zszn7mmWeIjY3lm2++ITc3l5ycHL744guGDRv2j1sC8VKzdmZmJt7e3hw+fJjz58+TlZXFjh07ir1n6NCh/Pjjj/Ts2ZPnnnuOn3/++ZpiBQcHs3btWiwWCxkZGWzduhWTyVSoP2RQUBBr1qwhNzeX/Px8VqxYQVBQEACOjo5XTCgB7rzzTk6dOsX27dsBWLFiBS+88EJpvg4RERExiNlismmrqEqsiXzggQd48cUXmTBhAocOHQLgjjvuYNy4cdx3332GFLC8XGrWHjhwIG5ubrRp04bQ0FD8/Py49957i73n6aefZvTo0cyePRtnZ2fi4uKuKVb37t3Zs2cP4eHhVK9enVq1alGlShV27dpl7fPYtm1bdu/eTdeuXcnPz6dVq1b07dsXgIceeoiIiAgSEhKKfb6LiwszZsxg4sSJnD9/Hg8PD2stq4iIiFQsFXmEtS1MlpImLLxMVlYWAF5eXnYtUGWwceNGLBYLbdu25fTp03Tu3JkVK1ZU2O/uqdsfNSyWV8l/t5S7HQUnDI1XYPCyBLc5eRoab++FPw2Nt/34L4bGO/fHJsNitWwywLBYAG1dahka740/vjY0Xpdbja3YOJpfvlPHXc0Jg+PVcfUxNF4ny82Gxnv68AeGxku7Pdym+5scXF1OJSlfV/2Nfu7cOXJycqhevbp11LGDgwN9+/bF0dHRiDL+4wQEBPDiiy/y5ptvAhATE1NhE0gRERGRsigxiTxw4AADBgwgOjqaPn36MG/ePFq0aMHu3btxd3fn0UeNq8H6J6lduzYffvjh9S6GiIiIVAAVuV+jLUocWPPGG2/wwgsvWKed8fT0ZNKkSbz++uskJiYaUkARERGRf7LKumJNiUnknj17CAsLs+5f6j5Zt25djhw5Yt+SiYiIiFQCN+Rk45cmwr5k+vTp1s/q4yciIiJydZW1ObvEJNLR0ZHs7Gw8PS+OAL3jjjsAOHnypAbVVCLPWM4bFiu1wPXqF5WjO0zVDY3nbvBfjOa8q19TnmZFuhsaz+TWwtB4Ro6YTklbYFgsgGVNxhkaL2tYoKHx3l9i7H+bfTucNTSeY936hsbL/eagofGWphk7OlvKR4nN2eHh4YwePZoLFy5YjxUUFDBhwgQ6d+5s98KJiIiI/NNV1j6RJdZEPvbYY7z44ouEhIRwzz33YDKZ+P777wkMDCx2jWcRERERKeyGbc5+44032LVrF9u3b8disTBw4MAia2mLiIiISPEq8NgYm5SYRObn5+Pk5ETDhg1p2LBhoXN79uyhfn1j+2iIiIiISMVQYp/IyycTHz9+fKFzI0eOtE+JRERERCoRs8Vk01ZRlZhEXr6s9nfffXfFczeq9PR06tWrR0pKSqHjISEhpKenX/G+qKioYo//+OOPjB49GoBly5bxySefADBjxgy++OKLciq1iIiIGOmGHFhjMl254CWdu5E4OzszduxYVq1ahYeHxzXds3Xr1mKPN27cmMaNGwMXk/YWLS5Ob/Lcc8+VT2FFRETEcObrXQA7ueaaSCmer68vwcHBvPbaa4WOb926tVCN44gRI0hISGDChAnAX10FgoKCeOKJJ4iIiCAlJYWoqCg2b97Mhg0bmDlzJps2bbLeC7BixQrCwsIIDw9nxIgRnDlzBoBWrVoxfvx4OnfuTNeuXTl8+LARry8iIiJXYcFk01ZRlZhEqrbx2owYMYJvvvmmSLN2ccaMGQPA8uXLgYsTt0dHR5OUlIST08WK4eDgYEJCQoiJieHf//639d69e/fyv//9j0WLFrF69Wrc3NyYNWsWAMeOHeOBBx4gMTGR+++/n8WLF5f3a4qIiIhYldicvWfPHu655x4AcnNzrZ8tFkuhCchvdB4eHowfP97arF1aTZs2vabrtm3bRtu2bfH29gagR48ehQY4XUo477rrLrZv317qcoiIiEj5MxvYsLt69WrmzJlDfn4+/fv3LzKv9+eff058fDwWiwV/f38mTZpEtWrVyhSrxCTy888/L9NDb0StWrUq1Kzt7+9fqDtAXt6V16erUqXKNcUwmwv3qrBYLOTn51v3XV0vLiloMpnUFUFERKSCMBvUJJ2RkcH06dNJSEjAxcWFnj17EhgYSN26dQHIyckhLi6OFStWUKNGDWbMmEF8fLy1lbS0SmzOXrJkCX5+flfcpLBLzdqZmZl4e3tz+PBhzp8/T1ZWFjt27LBe5+joWCj5K46joyMFBQWFjrVo0YINGzaQlZUFXBzBHRho7Pq0IiIiUjpG9YncvHkzQUFBeHl54e7uTvv27Vm/fr31fF5eHi+99BI1atQAoF69ehw5cqTM71ViTeTmzZvL/OAb0aVm7YEDB+Lm5kabNm0IDQ3Fz8+Pe++913rdQw89REREhHWwTHGCg4OZNm0aVatWtR6rX78+Tz31FFFRUeTl5dGwYUNefvllu76TiIiI2MbW0dnZ2dlkZ2cXOe7p6Ymnp6d1PzMzk+rVq1v3fX19SUtLs+57e3vzn//8B7jYTXHu3LlXnHbwWpSYRErJ/P392bBhQ6FjrVq1Yu/evQC88sorxd4XHx9v/XzpWoDAwEBrzWJoaCihoaEAdOjQwXrNo48+WmgS+OKeExkZSWRkZGlfR0RERCqghQsXWgfSXm7w4MEMGTLEum82mwsNirZYLMUOkj59+jSDBg2ifv36dOnSpczlKjGJ3L9/P+Hh4Vc8v3r16jIHFhEREbkR2DpNT//+/YtN9i6vhQSoWbNmoYG1x44dw9fXt9A1mZmZDBw4kKCgIEaNGmVTuUpMImvUqMHYsWNtCiAiIiJyI7O1OfvvzdZXEhwcTHx8PCdOnMDNzY1PP/200LLVBQUFPP300zzyyCM8++yzNpbqKknkTTfdZF01RURERERKz6gVa2rUqMGwYcPo168feXl5dOvWjSZNmhAdHU1MTAxHjx7l559/pqCggOTkZAAaNWrExIkTyxSvxCRS08SIiIiI/HOEh4cX6Yo4b9484OLyynv27Cm3WCVO8bNo0aJij+fm5rJkyZJyK4SIiIhIZVVZlz0ssSby8ull4OIklosWLWLZsmVUq1aN3r1727VwYozdeVWvflE5OeVsWCgAzmNsbbpDiX+WlT9Hg//nsirhZkPj5Rj8fbZ1ubaJ/8vDsibjDIsF0D2t+Nki7OXdZsa+30lHY/+tL02uYWg8t3UFV7+oHOWZahsa70+nyt3yaa64eaBNrmmKn7S0NN577z0+/fRTTCYTcXFxdO7c2d5lExEREfnHM2rFGqOV+Hf+unXr6NmzJ9HR0Xh7e7NkyRKqV69O165dcXR0NKqMIiIiIv9YFhu3iqrEmshhw4bRqVMn3nrrLW6++WIzVnGTVoqIiIjIjaXEmsi33nqLkydPEhISwtChQ9m4caNGbIuIiIiUgtnGraIqMYkMCQlh3rx5rF69Gj8/P0aOHElGRgazZs0iKyvLqDKKiIiI/GOZTSabtorqmsY+1q5dm9jYWL766iteffVVvvnmG9q2bWvvsomIiIj841XWPpElJpEXLlxg+PDhfPbZZwC4uLiwdu1abr/9dj788ENDCni9pKen06hRIyIiIujcuTOhoaEMGDCAo0ePEhISQnp6+vUuooiIiPwD3JDN2TNnziQnJ4d77rnHeuyVV14hOzubdevW2b1w15uvry9JSUkkJiayZs0a6tWrx5QpU653sURERESuuxKTyI0bN/LGG2/g4+NjPVajRg2mTJnC559/bvfCVTSBgYH88ssvAMyePZvOnTvTvn17fvjhBwC2bt1Kr1696NKlCw899JD1OxoxYoR1wfMNGzYQEhLCG2+8QWRkJN27d2fjxo3069ePNm3asHbtWgD27dtHVFQUXbt2pW3bttaa36ysLAYNGsQjjzxCREQEW7ZsYfny5Tz//PPWcsbHxzN37lwjvxoRERG5ArPJtq2iKjGJdHZ2pkqVois4eHh44OLiYrdCVUR5eXkkJyfTrFkzAOrWrUtiYiJRUVHMnz8fgA8++IAJEyawcuVKJkyYwIwZM6z3e3l5sW7dOkJCQgC45ZZbSEhIICAggLlz5/Luu+8ydepUa/K3fPlynn32WVasWMH7779vrQGdMWMGt912G+vWrWPKlCm8+eabdOzYkS1btpCTkwPAJ598QkREhGHfjYiIiFyZGZNNW0VV4jyRDg4O5OTk4OHhUeh4Tk4O+fn5di1YRZCZmWlNxi5cuECTJk14/vnnSUlJoV27dsDFZDI5ORmAqVOn8uWXX7J+/Xp++OEHzpw5Y31WkyZNCj27devWANSqVQtfX1+cnJyoVasW2dnZwMXay02bNvH222+zb98+zp49C8C2bdt4/fXXAahXrx5Lly4FoE2bNnz22WfUrl2b2rVrU6OGsUtyiYiISPEq8uAYW5SYRIaFhTFmzBheffVV3N3dATh79ixjxozh4YcfNqSA19OlPpHFubRiz+WTr/fu3ZvAwEACAwN54IEHeOGFF6zn/l6j6+z81yLSTk5FfwxDhw7F09OTtm3b0rFjRz755BPrtZfH/O2337jjjjvo2rUrc+bMwd/fn8jIyDK8rYiIiMi1K7E5u3///lStWpWWLVvSvXt3unXrRsuWLfH09GTQoEFGlfEfISsri4MHD/Lcc8/RunVrvvjiCwoKCsr8vJSUFGJiYmjXrh1ff/01AAUFBdx3332sWbMGuJhARkdHYzKZuO+++zh69CipqanWWlIRERG5/iprn8irNmePHz+ep59+ml27duHg4ECTJk3w9fU1qnz/GF5eXnTr1o3Q0FCcnJwICgoiNzfX2gxdWkOGDKF37964urpSv359/Pz8SE9PJyYmhjFjxtCpUyecnJyYMmWKtWbyP//5D1lZWTdcf1UREZGKrCJP02MLk0XrGP7jWSwW8vLyGDBgAKNGjaJhw4aluv/DWn3sVLKiDjsb+yfVeYN7olzT7P3lyNHgDte18wwNR47BX+h+p7K3HpRWkwuOhsUC6J72iqHx3m02ztB4xx2N/bfuW2Dsvz03g7OQPINrv9KdjP35jT202NB4C/z62nT/gN8/KKeSlC+jf+eJHRw7doyWLVvStGnTUieQIiIiYl83ZHO2/DP4+vqybdu2610MERERuYEoiRQRERGxo8raJ1JJpIiIiIgdKYmUSutOzhkW60+zu2GxAGoZ3Pndw2zs/yr+dDL2/SKiyjbbQFm5Pv+aofHc/B80NF7WsEDDYp2f+gKLFxv77+/xncYN5jF6IE/3tkcMjefc/E5D451c/puh8Rb/UcvQeEazVOB+jbZQEikich0YmUAClTqBFKnoKmtNpEZni4iIiEipqSZSRERExI4qa02kkkgRERERO6qsq7ooiRQRERGxo4o8Ybgt1CdSREREREpNSWQx0tPTadSoEREREXTu3JnQ0FAGDBjA0aNHAQgJCSE9Pd2mGNHR0WRkZJT5vIiIiPwzmG3cKio1Z1+Br68vSUlJ1v3JkyczZcoUpk2bVi7Pnzdvnk3nRURE5J+hIieCtlBN5DUKDAzkl19+se7Pnj2bzp070759e3744QcOHTrEgw8+iPn/TzadmprKE088wdGjR+nbty+RkZF069aNnTt3An/VZp4/f55Ro0bRvn17wsLCWLt2baHzOTk5xMTE0KNHD9q2bcuoUaOwWCykpqby+OOP8+yzz9K+fXtiYmK4cOGC8V+MiIiIlMhi41ZRKYm8Bnl5eSQnJ9OsWTPrsbp165KYmEhUVBTz58+nTp06+Pv7k5qaCkBiYiKRkZF8/PHHPPjggyQkJBATE8OOHTsKPXvRokWcPXuWdevWsWDBAmbPnl0oGdy4cSMNGjRg6dKlJCcns23bNnbt2gXA999/z7hx41i3bh1//PEH33zzjQHfhoiIiJSG2WTbVlGpOfsKMjMziYiIAODChQs0adKE559/3nq+Xbt2wMVkMjk5GYCuXbuyatUqmjVrxrfffktcXBw1atRgyJAh7N69mzZt2tC3b99CcbZt20b37t1xcHCgevXqrFmzptD5sLAw0tLSeO+999i/fz9ZWVmcPXtx6bm77rqLmjVrAhAQEMCpU6fs82WIiIiI/I2SyCv4e5/Iv3N0dATAZPrrT4QOHTowffp0kpOTad26Na6urtx7772sWbOGjRs3snbtWlauXMmCBQus9zg5ORV6xqFDh7j11lut+4sWLSI5OZnu3bsTHBzMvn37sFguVm67urparzOZTNbjIiIiUnGoT6RclZubG61bt2batGlERkYCMGXKFFatWkWXLl0YN24cP//8c6F77r//ftauXYvF/7lQTAAAIABJREFUYuHPP/+kb9++hZqzU1JS6NGjB506deL8+fPs2bPH2u9SREREKj71iZRrEhoaioeHB02bNgUgKiqK5ORkIiIiGDx4MK+99lqh63v37o27uzudOnXiscceY+zYsXh4eFjP9+/fn1mzZhEeHs6rr75K8+bNbZ5eSERERIxjxmLTVhqrV6+mY8eOPPzwwyxevLjI+d27dxMZGUn79u0ZPXo0+fn5ZX4vNWcXw9/fnw0bNlzx/OXnAgMDCQwMBKCgoICUlBQeffRR6/lbb72VJUuWlPiMV1555Yrn/f39rX0u/+5SXLg4BZGIiIjcuDIyMpg+fToJCQm4uLjQs2dPAgMDqVu3rvWa2NhYJkyYQLNmzRg1ahTLli2jd+/eZYqnmshy1LVrV3bt2kWvXr2ud1FERESkgjBqsvHNmzcTFBSEl5cX7u7utG/fnvXr11vP//777+Tm5lpnm4mMjCx0vrRUE1mOEhMTr3cRREREpIKxtV9jdnY22dnZRY57enri6elp3c/MzKR69erWfV9fX9LS0q54vnr16jatjqckUkRERMSObB0Ou3DhQmbNmlXk+ODBgxkyZMhfcczmQjO+WCyWQvtXO19aSiJFRERE7MjWCcP79+9Ply5dihy/vBYSoGbNmmzfvt26f+zYMXx9fQudP3bsmHX/+PHjhc6XlvpEioiIiFRgnp6e+Pv7F9n+nkQGBwezZcsWTpw4wblz5/j0009p3bq19byfnx+urq7W1fOSkpIKnS+t/8fevcf1fP//H7+9Owk5NMTIjJhsDstQRKa2ieTdEUPlMMacPnOaQ5jTlvOZYSZn5lQibJYZk4SZs/kMQ6hMNJIO79fvj369v6WcPt6vF+Nxdely6f16v3vdX++3Do/386hTZIXqV97XVTo//kEmkqnxilefva3tckg3/iyuad7F2yUf/yATOltE284La42XRN1l/o9mWS0MJTTLAkgx0/Znr2y2tnu1dTtacJULNX3jNFrTPCuN/1Kna9zEVDFT2ycYeK3g0jdqCn3zf5v9nGvCxYKrvDxMVFQUCxcuJDMzk4CAAHr06EGPHj3o378/derU4cyZM4SGhnLnzh3eeecdvv76a6ysrP6n65LubCGEEEIIFWlZInt7e+Pt7Z3v2OLFi42fOzo6smHDBpNkSREphBBCCKGil3WfOSkihRBCCCFU9LS7zvxbyMQaIYQQQgjx1KQlUgghhBBCRS9nO6QUkUIIIYQQqpIxka+gK1eu4OnpiYODA5Cz0vvdu3fx8fGhf//+T3SOOXPmANCvXz/0ej2RkZEPfeymTZs4ePAgYWFhz37xQgghhHghvKxjIqWIfAw7O7t8hV9iYiItW7bEy8vLWFw+qUcVkEIIIYQQ/yYyseYpJScnoygKxYsXJzQ0lPbt2+Ph4cFnn31Geno6AN9++y0fffQR7du3z7fxec2aNQG4d+8egwYNok2bNnh7exMREVEg5+jRowQGBtK2bVtCQkL466+/APjjjz/w8/NDr9czfvx4PvzwQ+7cuYOzszN37twBclpQW7durfZLIYQQQognoDzjx4tKWiIfIykpCb1ez/3790lJSaFOnTrMnTuXy5cvY2lpybp16zAYDISEhLBnzx4qVqzIxo0b2bx5Mzqdjvbt21O3bt1855wzZw62trZs3bqVmzdvEhgYiKOjo/H+jIwMBg4cyMyZM6lbty7bt29n4MCBbNy4kWHDhjFgwACaN29OeHg42dnZ2NjY8P7777Njxw4CAgKIiIjAx8dH65dKCCGEEIV4WcdESkvkY+R2Z0dHR6PX61EUBVdXVxo2bEjHjh1ZtWoVEydO5OLFi6SlpXHw4EGaN29O8eLFKVasGJ6engXOeeDAAQICAgB47bXX8PDw4ODBg8b7L168SMmSJY3FZ6tWrbh06RIJCQkkJCTQvHlzAPz9/Y1f4+/vb+wu37p1K3q9XrXXRAghhBBPTnnGfy8qKSKfkJmZGUOHDiUxMZElS5bw008/MXjwYKytrfHz86Nhw4YoioJOpyPvduQWFgUbex/crlxRFLKzs423DYaC71lyv+ZhW503bNiQpKQkfvjhB+zt7Slfvvz/9DyFEEIIYVqGZ/x4UUkR+RQsLCwYOnQo8+fP5+eff6ZVq1b4+/tTsmRJ4uLiyM7OpnHjxuzevZt//vmH+/fv8+OPPxY4j4uLi3Hfyps3b/LTTz/RqFEj4/3VqlXj1q1bxvGU0dHRVKxYkUqVKlG5cmX27NkD5Gyynkun0+Hj48OECRPw8/NT82UQQgghhJAxkU/Lzc0NJycnLl68yNGjR9m2bRuWlpbUr1+fK1euEBgYSEhICAEBAZQsWZKKFSsWOEefPn348ssv8fb2Jjs7m169evHOO+9w9uxZAKysrJgxYwbjx4/n3r17lCpVihkzZgAwefJkRowYwcyZM6lZsybW1tbG83p5efHdd9/xwQcfaPNiCCGEEOKxZImfV5C9vT0xMTEFjn/33XeP/LpOnTrRqVOnAsdzi0QbGxumTp1a4H4/Pz9jK6KTkxPr168v8Jjt27czZ84c7Ozs+OGHH7h79y6Q0wW+d+9e9Ho9VlZWj39yQgghhNDEy1lCShH5r1OxYkW6deuGhYUFJUuWZOLEiQD07duXa9eusWTJkud8hUIIIYTIS1oixQshb2tlXvPnz38OVyOEEEKIx3mRJ8c8C5lYI4QQQgghnpq0RAohhBBCqOhFXuvxWUgRKXBKz9Is679W2n7LrT1RWdO8ohr/njDX+Ce4/ZtXNM2zrqxtZ8nKn3WaZXX2TNMsC2DdTm3Xjm3X4pqmed84jdY0r9dv4zTNS/u8h6Z5iYcsNc3blFZW0zytvazd2VJECiGEEEKo6GVtiZQxkUIIIYQQ4qlJS6QQQgghhIqkO1sIIYQQQjw1g/JydmdLESmEEEIIoaKXs4SUIlIIIYQQQlUv6441/4qJNVeuXKF27dro9Xr0ej3e3t64u7sze/bsx36du7v7U2W5u7tz5cqjlzHJyMhg7NixtGnTBm9vbzp16sSxY8eeKudhevToQWJioknOJYQQQgihln9NS6SdnR2RkZHG24mJibRs2RIvLy8cHBw0vZbw8HAMBgNRUVHodDoOHz7MZ599xu7du7G0fLa1tRYvXmyiqxRCCCHEi+BlXeLnX1NEPig5ORlFUShevDgAixYtYvv27WRnZ9O0aVOGDBkCQHp6Op9//jnnzp2jZMmSzJs3D1tbW1auXElkZCT37t3D0tKSadOmUa1aNQDmzp3LmTNnKFKkCGPHjsXR0TFf9o0bN8jMzCQzMxMrKyvee+89vvrqKwwGw0OvJSEhgU8++QRbW1usra25desW48ePp3bt2mRnZ9OiRQs2b95MYGAgy5cvp1y5cowdO5bDhw9jaWnJZ599RuvWrTl27Bhff/016enp2NraMnbsWCpXrszSpUvZvHkzZmZm1K1bl3HjtF0IVwghhBCFe1lnZ/8rurMBkpKS0Ov1eHp64uzszMyZM5k7dy4VKlTgl19+4cSJE2zYsIGIiAgSExPZsmULADdv3qRr165s3bqVsmXLEh0dzZ07d9i1axcrVqxg69atvP/++6xatcqYVaVKFSIiIvjss88YNmxYgWsJDg7m999/p3HjxvTu3Zvly5fj5OREkSJFHnktFy5cYMqUKSxduhS9Xs+2bdsAOHDgAI6OjpQpU8aYsWLFCtLS0ti+fTtLly5l3rx5ZGRkEBoayrRp09i8eTNdu3Zl1KhRZGdns3DhQjZu3MimTZvIzMyULnEhhBDiBWFAeaaPF9W/piUytzvbYDAQFhbGn3/+iaurKwCxsbEcO3YMPz8/IKf1sWLFirz33nvY2dlRt25dAKpXr05KSgo2NjZMmzaNbdu2cfHiRfbu3UutWrWMWYGBgQA0b96cIUOGkJqaSsmSJY3329vbs3XrVo4fP87+/fuJiIggPDyciIiIR15LmTJlsLe3B8DLy4v27dszdOhQtm7dStu2bfM93/j4eNq1a4eZmRnlypVj27Zt/PHHH1y+fJnevXsbH3fnzh3Mzc1xcnIiICAADw8PunbtSvny2m5xJoQQQojCSXf2C8LMzIyhQ4fi4+PDkiVL6NGjB9nZ2YSEhNC1a1cAUlNTMTc3JyUlBQuL/3uKOp0ORVG4du0aQUFBdO7cGTc3N8qWLcvp06eNjzM3Nzd+rihKvnMATJ8+nU6dOlG3bl3q1q1Lr1696NChA7/++usjr8Xa2tp4jnLlylG1alXi4uKIjY1l9Oj8+75aWFig0/3fPr5//fUXBoMBe3t749jQ7Oxsbty4AcD8+fM5evQov/zyC5988glTp06lUaNGz/RaCyGEEEI8zL+mOzsvCwsLhg4dyvz580lOTsbFxYXIyEju3r1LVlYWffr0YefOnQ/9+uPHj1OlShW6dOlCnTp12LVrF9nZ2cb7o6KiAPjxxx9xcHCgWLFi+b4+MTHR2L0MOeMzb968yVtvvfVU16LX65k0aRLOzs4ULVo0330NGzYkOjoaRVH4+++/6dy5M5UqVeL27dscOnQIgI0bNzJ48GBu3rxJ69ateeuttxgwYACurq6cPXv26V9YIYQQQpic4Rk/XlT/upbIXG5ubjg5OTFr1iwmTJjAmTNnaNeuHdnZ2TRr1gxfX18SEhIK/VpXV1fWrFlD69atURSFhg0bcu7cOeP9Fy9eRK/XU7x4ccLCwgp8/ahRo5g0aRKenp4ULVoUS0tLBg8ejIODAw4ODk98LR9++CFjxoxh8ODBBe7r2LEjEyZMMHZzjxo1ihIlSjBr1iwmTpzI/fv3sbGxYdKkSbz22mu0b9+egIAAihYtStWqVfH39/9fX1ohhBBCmJDyku5Yo1Ne1mcmntiO8h00y/qv1b/2fcsTKarxT5O5xnmeDo9eQ9XUrCtr21ni/bPu8Q8ykW2e2j63dTu1HSfdrsU1TfOW//y6pnm9ftN2BYy0z3tompd46NmWq3tam9LKapo37K+Vmubp32jzTF8feWmria7EtP6V3dlCCCGEEOL5kiJSCCGEEEJFz3tM5NWrV+nUqROenp707t2bu3fvFnhMUlIS3bt3R6/X4+vrS2xs7GPPK0WkEEIIIYSKlGf896zGjh1Lx44d2bFjB7Vr12b+/PkFHjN58mTc3d2JjIxk2rRpDB48ON+k48JIESmEEEIIoaLnudh4ZmYm8fHxtGzZEgA/Pz927NhR4HEffvghbdrkjN2sUqUK9+/fJy0t7ZHnfrlnOQghhBBCPGfPOoc5NTWV1NTUAsdLliyZbzOUwuRuspK75nW5cuUK3dUut8gEWLJkCbVq1aJEiRKPPLcUkYLiuizNsiw0/pbTaTx7uYjGC3ppN5c4R7FaRTTNs6j5hqZ5N3cd0yzLvLqjZlkARbc/ulvK1CydqmmaZ7X7nqZ5Ws+WLjZjsaZ5Nm27a5pnWXCInshj2bJlzJ07t8Dxvn370q9fP+Pt7du38/XXX+d7TJUqVfJtXgIUuJ1XeHg469atY+XKx89glyJSCCGEEEJFz9q+EBISgq+vb4HjD7ZCtmrVilatWuU7lpmZibOzM9nZ2Zibm5OcnIydnV2hOZMnT2bPnj2sWrWKChUqPPa6pIgUQgghhFDRs06OeZJu64extLSkQYMGREdH4+3tTUREBG5ubgUeFx4eTlxcHGvWrHniLCkihRBCCCFU9KyTY57VmDFjGDZsGAsWLOD1119n+vTpAKxZs4akpCT69+/PvHnzsLGxISgoyPh1ixYtonz5h29UIEWkEEIIIYSKnvfmgJUqVWLFihUFjn/88cfGz+Pj45/6vLLEjxBCCCGEeGqvXBF55coVateujV6vR6/X4+3tjbu7O7NnzzZpzqxZs/jpp5+e+PH//PMPffr0Mek1CCGEEOL5e57rRKrplezOtrOzIzIy0ng7MTGRli1b4uXlhYODg0kyBgwY8FSPv337NqdPnzZJthBCCCFeHKbYdeZF9EoWkQ9KTk5GURSKFy8O5Awk3b59O9nZ2TRt2pQhQ4Zw9+5dBg4cyI0bNwDo06cPHh4eBAUF4ejoyKFDh7h//z4jRoygadOmDBs2jEaNGtGoUSM++eQTbG1tsba2Zs6cOYwYMYLExESSkpJo3LgxEydOZMKECSQlJdGnTx/mzZvHxo0bWbp0KTqdjnfeeYdRo0ZRvHhxXFxcqF27NsnJyTg4OODi4kK7du0ACAoKYvDgwdSrV++5vZZCCCGEyM/wnMdEquWVLCKTkpLQ6/Xcv3+flJQU6tSpw9y5c6lQoQK//PILJ06cYMOGDeh0OoYMGcKWLVswGAxUqlSJRYsWcfr0abZs2YKHhwcAd+7cYfPmzZw+fZoePXoQExOTL+/ChQt8++232Nvbs3XrVmrVqsXs2bPJyMjAy8uLkydPEhoaSnBwMPPmzePs2bN88803fP/999ja2jJ27Fjmzp3LF198QUpKCj169MDZ2ZkDBw4wZ84c2rVrR0JCAjdv3pQCUgghhBCaeCWLyNzubIPBQFhYGH/++Seurq4AxMbGcuzYMfz8/ABIT0+nYsWK+Pv7M336dBITE3n//ffzjV/MbQmsVasW5cqV4+zZs/nyypQpg729PQBt2rTh2LFjhIeHc/78eW7dukVaWhqlS5c2Pj4+Pp4WLVpga2sLQPv27Rk+fLjx/txC0dnZmVGjRnHlyhUiIyPR6/WmfqmEEEII8YxeznbIV7SIzGVmZsbQoUPx8fFhyZIl9OjRg+zsbEJCQujatSuQs1+lubk5xYsXZ/v27ezdu5fdu3fz3XffER0dDYC5ubnxnAaDwbg/ZS5ra2vj5ytWrGDnzp20a9eOJk2a8McffxSY+m8w5F/bXlEUsrL+b2vC3PPpdDp8fHzYtm0b27dvZ8mSJSZ4VYQQQghhSi/y5Jhn8crNzn6QhYUFQ4cOZf78+SQnJ+Pi4kJkZCR3794lKyuLPn36sHPnTlauXMmcOXNo1aoVY8aM4ebNm9y5cwfAWEweP36c1NRU3nrrrYfm/frrr7Rv3562bdty//59zpw5Yyw8cwvFRo0aERMTw61btwD4/vvvcXZ2LvR8fn5+rF27ltdff/2RC4IKIYQQ4vmQ2dkvMTc3N5ycnJg1axYTJkzgzJkztGvXjuzsbJo1a4avr69xYo23tzfm5uYMGTLEuC3Q5cuXjXtazpgxI1/L5INCQkL48ssvWbRoETY2Njg5OXHlyhUaNGhAxYoVCQoKYsWKFXz66acEBQWRmZnJO++8w9ixYws93+uvv87rr79e6J6aQgghhHj+nvdi42p55YpIe3v7AhNfAL777jvj55999hmfffZZvvttbGxYtGhRoecMDg4u0FIYFhZm/DxvXuPGjdm5c2eh51m7dq3x88DAQAIDAws8Ju94S0VRSEpK4saNG3zwwQeFnlMIIYQQQg2vfHf2v9nOnTvR6/UMHDgQKyur5305QgghhCiEdGeLQhW2F6VWPD098fT0fG75QgghhHg8WWxcCCGEEEI8NRkTKYQQQgghntqL3CX9LGRMpBBCCCGEeGrSEinIUB6+JJGpZWqWlCND47dJGTpt87I0zrt7MkPTPMOxPzXNq1KkjGZZ6fsuapYFkKmrrGleynpt/+/SzSpqmpd4yFLTPJu23TXNK7dF480r6o/WNk9j0p0thBBCCCGe2svanS1FpBBCCCGEil7W2dkyJlIIIYQQQjw1aYkUQgghhFCRQcZECiGEEEKIp/WydmdLESmEEEIIoaKXtSXylRwTuWPHDvz8/Gjbti3e3t58++23/9N5jh8/zsiRIx/5mGHDhrFp06Z8x0JDQwkPDzfeXrlyJTVr1iQxMdF4rH379sTFxT3xtbi7u3PlypUnfrwQQgghtKE8478X1StXRCYmJjJp0iSWLFnCli1bWLt2LdHR0fz0009Pfa46deowceLEp/46FxcXjhw5Yry9b98+mjZtyt69ewFIT0/n/PnzODk5PfW5hRBCCCG08MoVkSkpKWRmZpKeng5A8eLFCQsLo3r16gAcPXqUwMBA2rZtS0hICH/99RcAp0+fJjAwEG9vbzp37sz169eJi4sjKCgIgIMHD/Lxxx/j6+uLh4cHu3bteug1uLi48NtvvwGQkZHBn3/+SUhICPv27TNeg5OTE1ZWVly4cIGgoCC8vb1p3749x44dA3JaOHv16kWrVq2IiYkxnvvChQt89NFHHD161MSvnBBCCCH+FwZFeaaPF9UrV0Q6Ojri4eHBBx98QEBAAFOmTMFgMFClShUyMjIYOHAgo0aNYsuWLXTo0IGBAwcCMHjwYD777DOioqJo3bo1y5Yty3felStXMmHCBDZv3syECROYNWvWQ6+hbNmylCpVisuXL3PkyBHeffddGjVqxOHDhzEYDBw6dIgmTZoAMGTIEIKCgoiKimL48OEMGDCAjIycXUNKly7N9u3bcXd3B+D69ev07duXr776infffVeNl08IIYQQT0m6s18iY8eOJSYmho8//pirV6/Srl07fvjhBy5evEjJkiWpW7cuAK1ateLSpUskJCSQnJxMixYtAOjYsSNffPFFvnNOmTKFc+fOMW/ePJYuXcrdu3cfeQ25Xdr79u3D1dUVa2trHBwcOHv2LIcOHaJp06bcvXuXS5cu8dFHHwHw7rvvUqpUKc6fPw9gvM5cAwYMoHLlyjRo0MAkr5MQQgghnp20RL4kfv75Z6Kjoylfvjz+/v7MmDGD0NBQNmzYgMFgKPD43P0udbr/26T4/v37XL58Od/jOnbsyLFjx6hduza9evV67HW4uLhw9OhR9u/fj6urKwCurq4cPnyYa9euUb169UL32lQUhezsbACsra3z3Tdy5EguX77Mzz///Nh8IYQQQmhDWiJfEtbW1kybNs04k1lRFE6fPk2tWrWoVq0at27dMo47jI6OpmLFilSqVIny5csbxyxGRkbm666+desWFy9eZMCAAbi5ufHTTz8ZC72HcXZ2Jj4+HkVRKFeuHJBTRK5bt4769esDYGNjg729PT/88AOQM1byxo0b1KhRo9Bz1q1bly+//JJx48aRlpb2DK+SEEIIIcSjvXLrRLq4uNC3b1969epFZmYmAM2aNaNPnz5YWVkxY8YMxo8fz7179yhVqhQzZswAcrqrv/zyS6ZMmYKtrS2TJ0/mwoULQM7YxICAALy8vLCwsMDFxYX09PRHFnIlSpSgaNGi+bqea9Wqxd9//20cD5k3d86cOVhaWjJnzhysrKweet6GDRvi7OzMzJkzGTFixDO9VkIIIYR4dopSsKfzZaBTCuszFa+Un8q31yzrlJWlZlkAGbrHP8aUrDT+acrS+Pl1eCNB0zxDtrZPsEeC9eMfZCIr6jx63LSpbTlaWdO8j964qmnehoSKmua1Kn5D0zwb2/ua5pXbskTTvNn1R2uaN+jSSk3zqpSp+/gHPcJffx8z0ZWY1ivXEimEEEIIoaWXtb3ulRsTKYQQQgghnp20RAohhBBCqMjwAs+wfhZSRAohhBBCqOhl7c6WiTWCWW901izrjk7bb7f30h+91JKplS9yT9O8SxnFNc27bqHtCJjb5prGUVLDCZRaz9X820zbn70iaDspqmqGts/vTyttn5/lS/6Xuv+RcZrmWZatpmne66Xffqavv3brlImuxLRkTKQQQgghhIqe92LjV69epVOnTnh6etK7d+9H7qp3584dPvjgA+Li4h57XikihRBCCCFeYmPHjqVjx47s2LGD2rVrM3/+/Ic+dvz48aSmpj7ReaWIFEIIIYRQkaIoz/TxLDIzM4mPj6dly5YA+Pn5sWPHjkIfGx0dTfHixalZs+YTnVsm1gghhBBCqOhZZ2enpqYW2jpYsmRJSpYs+civTUlJwcbGBguLnJKvXLlyJCYmFnjc1atXWbZsGcuWLaNHjx5PdF1SRAohhBBCqOhZWxOXLVvG3LlzCxzv27cv/fr1M97evn07X3/9db7HVKlSBZ0u/0SwB28bDAZGjhzJqFGjsLZ+8p27pIgUQgghhHiBhYSE4OvrW+D4g62QrVq1olWrVvmOZWZm4uzsTHZ2Nubm5iQnJ2NnZ5fvMefPn+f8+fOMHDkSgEuXLhEaGsr48eNxcXF56HW9ckXklStX8PT0xMHBAYD09HTq16/PoEGDSE9Pz3dfrnbt2tGpUycURSE8PJyIiAgAzMzM+OSTT/Dy8gLA3d0da2trLC3/b3/ot99+O9+7gn79+nHx4kWioqLyZaxatYrvv/8eRVHQ6XR07doVHx+fx5532LBhHDhwgFKlSmEwGLCwsKBHjx60bt3ahK+aEEIIIf5XhmdsiXySbuuHsbS0pEGDBkRHR+Pt7U1ERARubm75HlO9enX27NljvB0UFETfvn1xdnZ+5LlfuSISwM7OjsjISCCniXn69On079+fyZMn57vvQTNmzODUqVOsXLmSEiVKcP36dTp37oytrS1NmjQBYNGiRdjb2xf69Tdv3uTUqVOUK1eOI0eOUL9+fQB+//131q9fz7p167C2tubvv//G398fR0dHHB0dH3ve/v374+fnB8Dly5fp2LEjpUuXNl6TEEIIIZ6f570k95gxYxg2bBgLFizg9ddfZ/r06QCsWbOGpKQkBgwY8D+d95UsIvPS6XT069cPV1dX7ty589DH3b17l2XLlrFlyxZKlCgBQIUKFZg+fTpFixZ9oqyoqCgaNmzIW2+9xdq1a41FZHJyMoqicO/ePaytrSlTpgyzZ8/G1tb2qZ9P5cqVCQ4OZvXq1VJECiGEEC+A573tYaVKlVixYkWB4x9//HGhjy/ssYV55YtIACsrK6pUqcLevXtJSkpCr9fnu3/y5MlkZGRgYWFBlSpV8t1Xt27dfLd79uyZr9s5ODgYf39/ADZt2sTAgQN56623mDVrFiNGjKB06dK4ubmxadMmmjVrxrvvvouj716EAAAgAElEQVSzszN6vZ7y5cs/0Xkf9NZbb7F58+b/7cUQQgghhEk975ZItUgR+f/pdDqsra0f2p198uRJrKysHnueh3U7nz59muvXr9OkSRMsLS2pVasWERERdOnSBSsrK+bPn89ff/3Fvn372Lt3L0uWLCE8PJx33333ked9mKeZXSWEEEII8bRksXEgIyODCxcu0Lx584c+xsHBgfT0dK5evZrv+LZt21i2bNljMzZu3EhGRgYtW7bE3d2dCxcusHbtWgAiIiKIjY2lSpUqdOrUiW+++YaQkJCHjs18nLNnzxaYHCSEEEKI58OgKM/08aJ65YtIg8HAnDlzqFevHmZmD385rK2t6dSpE19++aVx7OSVK1eYPn36Ywu2jIwMoqKiCA8PJyYmhpiYGH766SeSk5OJi4sjOzubadOmcfPmTePjz507x9tvP/2G7RcvXmT16tUPHecghBBCCG09772z1fJKdmfnHfdoMBioVasW06dP559//il0TGTDhg0JDQ3l888/Z968ebRr1w4LCwvMzc0ZNGgQTZs2NT72wbGLRYsWpUuXLlSqVIl69eoZj9vY2BAYGMjatWuZMWMGKSkpfPzxx8ZC1svLi4CAgEeeN7clc/bs2SxbtgydToe5uTlffPGFcdKOEEIIIZ6vF7k18VnolJd1tKd4YrPe6KxZ1h2dtt9u76Vna5pXvsg9TfMuZRTXNO+6hbadF7fNNY2jpEG7LA2jAPjbTNufvSLoHv8gE6qaoe3z+9NK2+dn+ZL/pe5/ZJymeZZlq2maZ239xjN9fXr6JRNdiWm98t3ZQgghhBDi6b2S3dlCCCGEEFp5kcc1PgspIoUQQgghVPSyjhyUIlIIIYQQQkUvaxEpYyKFEEIIIcRTk9nZQgghhBDiqUlLpBBCCCGEeGpSRAohhBBCiKcmRaQQQgghhHhqUkQKIYQQQoinJkWkEEIIIYR4alJECiGEEEKIpyZFpBBCCCGEeGpSRAohhBBCiKcmRaQQQgghhHhqUkQKIYQQQoinJkWkeGHdvn27wLGEhARVsnr06MH27dvJyMhQ5fzi5XPmzJnnfQniXyIpKQmAQ4cOsWrVKtLT05/zFZnOsWPH8t1OT08nLCzsOV2N0JrsnS1eONeuXUNRFHr27MnixYvJ/RbNzs6mR48e7Nixw+SZBw8eJCIiggMHDtC8eXN8fX2pW7euyXMKk5iYSHZ2Nubm5pQvX17VrD179nDgwAGysrJwdnbmgw8+MHnG8OHDH3n/119/bfJMgPj4+Hy3dTodRYoUoUqVKpQsWdLkea1atWL79u0mP+/DJCQksHLlSm7fvk3eX9tqvZ7ffvster2ecuXKqXL+58nLywtfX19Nnt+YMWPIzMykW7dudO/eHVdXVzIyMpg6daqquVpp2bIlYWFhODk5sWfPHsaOHYuLiwtfffWVKnmxsbGsWbOG8+fPU6RIEapXr07Hjh2pV6+eKnni0aSIFC+c4cOHExcXR1JSEnZ2dsbjFhYWvP/++4wYMUK17PT0dHbs2MHMmTOxsbEhICCAjh07YmVlZbKMO3fuEBoaSp06dejevTtubm5YWFiQmprK3LlzcXFxMVlWXosXL+aHH37A29sbRVGIiorCw8OD3r17mzRn8+bNj7zf19fXpHm5unTpwokTJ2jcuDGKonDw4EEqVarEnTt3GDBgAG3atDFpXr9+/ahZsyb16tXD2traeLxhw4YmzckVGBhIgwYNqFGjBjqdznhcrddz7ty5bN26lTfeeANfX18++OADLC0tVcnSWkJCAhEREWzdupXKlSvj5+eHh4eHKs/Pz8+PjRs3MnfuXCDn+8bf35+NGzeaPAtyWga1egMMcOHCBfr160flypW5cuUKY8aMoUGDBqpkRUdHExYWRnBwsPHn4OzZs6xZs4Zhw4bx0UcfqZIrHkER4gn17du3wLHg4GDV8hYuXKjauQtz4MABZfjw4Yqbm5sSGhqq7N+/X5k6darSrVs3k+aMHDlSmTx5spKVlaUoiqLo9XpFURQlPj5e6dmzp0mz8mrTpo1y79494+20tDTF09NTtTxFUZSUlBTl6tWrSkJCgnLp0iVl//79qmUFBwcrCQkJxtvXr19XunXrpvzzzz+Kj4+PyfM6d+5c4CMoKMjkObnUeA5PIj4+XhkzZozSqlUrZezYscqpU6dMnpGZmans3r1b2bx5c74PLfzwww+Km5ub0qhRI2XChAnKzZs3TXr+tm3bKllZWYper1eOHj2qpKWlKa1atTJpRl6dO3dW2rRpoyxevFhJSkpSLSchIcH4ER8fr7i6uirbt283HlODr6+vcunSpQLHL1y4oAQEBKiSKR7N4nkXseLF17dvX06fPk1SUhIeHh7G49nZ2VSoUMHkebnv2B/8PO/1mFqLFi2wt7fH39+f0aNHG1uWnJ2d8ff3N2nWwYMH+eGHHwocb9CgAcOGDTNpVl6KouRrMStSpAgWFur9CpgzZw7h4eFkZWVha2tLYmIitWvXZv369arkJSUlUbFiRePt8uXLk5SUhI2NTb7uX1NZsWKFyc/5KO+99x4xMTE0bdrUpC3jj5KWlsaVK1e4fPkyZmZmlCpViokTJ+Lk5MSgQYNMljNo0CCuXr2Kg4NDvlZWHx8fk2XkdffuXXbu3ElkZCSJiYl8/PHHeHl58csvv9C9e3c2bdpksiwfHx+aNm1K/fr1qVevHq1bt6Z9+/YmO/+DVqxYQUJCApGRkXTr1o2KFSvi6+tr8pbWzp07G/+vFEXBysqKyZMnAzlDSX766SeTZeXKzMykcuXKBY6/+eabZGVlmTxPPJ4UkeKxwsLCuHXrFhMnTiQ0NNR43MLCgjJlyjzHKzOd7t2707lz5wLHzczMHts9+7Qe/EU+b9484+c2NjYmzcrLxcWFfv36Gbs/IyIicHZ2Vi1v8+bN7Nmzh4kTJ9K7d2/Onz/P6tWrVcurX78+gwYNwtvbG4PBwLZt23BycuLnn3+mWLFiJs9LSEggNDSUhIQEVq1axaBBg/jqq6+wt7c3eRbAjh07WLlyZb5jOp2O06dPq5I3ePBgYmNjad68Ob179zZ2UWZkZNC0aVOTFpFnz55VZazzw3h4eNCiRQv69u2bb/hBx44d2b9/v0mzunbtSkhICGZmOfNYV65cyWuvvWbSjAdVqlQJHx8fLCwsWLt2LStWrGDGjBkMHjyYDz/80CQZMTExAOzevZsWLVqY5JyPo+abXvG/kf8R8Vg2NjbY2NiwYMECzp07l29g/6VLl0w+BkyNlsbHWbt2baFFpBqKFSvGxYsXefPNN4GcX/gA58+fV6XYyTVy5EhWr15NREQEiqLg4uKiaouInZ0dNjY21KhRgzNnzvDRRx8xbdo01fLGjh3LmjVrWLduHebm5jRu3Jj27dvz66+/GltITGn06NF0796dqVOnUrZsWdq0acMXX3zBqlWrTJ4FsG/fPlXO+zAuLi6MHz+eokWL5jtuZWXFtm3bTJrl4OBQYAy0mnbt2lXoGzadTpfvTZ0p5H2zsXLlSgYPHqzqm43169cTGRlJcnIyPj4+rF69mgoVKpCYmIivr6/JishcU6dO1ayIvHXrFhEREQWOK4pS6GoeQn1SRIonNm7cOGJiYvJ1J+h0OpYvX65KnqOjY76uLcgpTPbs2WPyrAoVKhAcHEy9evUoUqSI8bgaBW23bt3o3bs3I0aMoEGDBuh0Oo4cOcKECRMYOnSoyfNy6XQ6fH19cXd3N74JeLAL2JRsbGyIiIjgnXfeYeXKldjZ2am6tImFhYVxAkje59e8eXNV8lJSUmjatClTp05Fp9PRrl071QpIgHv37jF37lxiY2PJzs7GxcWFAQMGqPbGw97enq5du7J27VrOnz9Pjx49mDJlCvXr1zf5jOb09HQ8PT1566238nXVq/W7xc/Pj+zsbONtnU6HtbU11apV44svvjC+sTOFvG82ypUrp/qbjQMHDtCvX78CvQzly5dnzJgxJs+rXLkyw4cPLzDBTI2hCM7OzsTFxT30PqE9KSLFE9u3bx87duzI94tCTXnX4cvMzGTXrl0cPXpUlax3331XlfMWplWrVmRlZTFhwgQuXboE5PwiHjBgAO+//75quXPnzmXJkiXY2tqi0+lQFEW1sUsAEydOZNu2bfj4+LB7925Gjx7Nf/7zH1WyAL755hsWLVpE6dKlNXl+1tbWXL9+3fhG59ChQ6qOVRw3bhxFixY1Lp3y/fffM2bMGKZMmaJK3qRJk5g0aRIA1apVY9GiRQwdOlSVWcWffvqpyc/5KG5ubtjb2xMQEADAli1bOH78OO7u7owcOZLw8HCTZWn9ZuP8+fMPLahatmxp8jxbW1sAfv/993zH1SgiZf3JF9DzmtEj/n26deumpKWlPddraNu2rSY5BoOh0FmApnbr1i3l1q1bqucoiqK0aNHC5DNPH2XYsGGaZSmKonh4eCh///23ZnnHjh1T2rZtq7z77rtK27ZtlebNmyu//fabanne3t4Fjqk5y7ewc6v583fy5EklPj5eOXjwoBIbG6usX79etazCZrr7+vo+9L5n8fHHHyvXrl0znjc+Pl7VmcSffPKJEh8fr9y/f1+1jAdlZGQoZ8+eVU6dOqVkZmaqljN8+HDj55s2bcp3X4cOHVTLFQ8nLZHiiZUqVQovLy+cnJzytbiotdhx3rEviqJw7tw51QZWr1u3jkmTJnHv3j3jMXt7e3788UdV8mJjY7Gzs8PBwQHImVFZvXp1GjdurEoe5AwFKFGihGrnf9Aff/zB3bt3KV68uCZ5r7/+OqVKldIkC6BOnTps2LCBixcvkp2dTbVq1VRtiVQUhdTUVOPC6ampqZibm6uWV61aNaZMmYJer0en07F161bjOF5TCw0N5eDBg9y+fZtq1apx5swZ6tevb2wpNDUzMzP27t1Ls2bNANi7dy9WVlbcuHHD5LN8hw0bxqeffsqlS5fQ6/Xcvn2bWbNmmTQjr+PHjxcY363mBKwTJ07Qv39/SpcujcFg4MaNG8ybN0+Vxb9PnTpl/Hz58uX51kjN+7tbaEeKSPHEmjVrZvylq4UHx77Y2toyc+ZMVbIWLlxIZGQkM2fO5PPPP2fPnj0cOXJElazo6GhmzJjBjBkzjMfKlCnDqFGjGDJkiMm7nHKXSSpZsiTt27fHzc0tX/Gh1kQmMzMzWrRoQdWqVfONM1VrnNubb75Jx44dcXZ2zlfMqfX8Ll++zNq1a0lJSdFkB5kuXboQEBBgHNO6e/duevbsqUoW5AxHmDlzJoMGDcLCwoIGDRowYcIEVbL279/Pzp07GT9+PMHBwdy7d0/VrsuwsDC++OILBg8eDMAbb7xBWFgY69ato1u3bibNqlu3br43G/b29qquwnDgwAHVzl2YCRMmMGPGDGPRePToUcaPH8+GDRtUzVUeWLbrwfHzQhtSRIon5uvry5UrV/jvf/9L06ZNuXbtWqFrdplK3j/Gd+7cUTWvTJkyVK5cmZo1a/LHH3/QqVMn1qxZo0rWt99+y8qVK/Ntcdi6dWvq1q1L//79VRm3BGi6iwXAkCFDNM0rX7686ttG5tWvXz8aN25snBylNn9/f+rUqUN8fDwGg4E5c+ZQs2ZN1fJKlSqVbyKGoihcuXJFldZsOzs7LC0tcXBw4OzZs3h5efHPP/+YPCdXjRo12LRpE1evXkWn0/H6668D0KdPH5NnRUdHs2DBAqKiorh06RJeXl6MGjVKlS1HQfsJWGlpaflaHd99913u37+vSlbenzMpGl8MUkSKJ5b7yzA9PZ21a9fSoUMHhg4dil6vVyVv/fr1HD58mKFDh+Lj40Px4sXR6/X06tXL5FlFixblwIED1KxZk127dlGnTh3VZhIrilJosWNvb4/BYDB5Xm5LXFZWFnv27MHDw4ObN28SExNj8oXU82rUqBGHDx/mjz/+wN/fn99//121LQFB+6WhFEXhiy++UD0ndx2+3OEducMDTp8+zenTp1VbkHvt2rVMnjw5XzdhpUqV2LVrl8mzypcvz8KFC2ncuLFxolBGRobJc3JdunSJgQMHcvnyZRRFoWLFisyYMYOqVauaPGvBggUsXboUyGnx3LRpE926dVOtiNR6AlapUqXYtWuX8fns2rWL0qVLq5KVmZnJtWvXMBgMxs9zWyQzMzNVyRSPZva8L0D8eyxevJg1a9ZQvHhxypQpw+bNm1m0aJFqeWvWrGHgwIFs3boVDw8PoqKiCt3pxRRGjRpFTEwMzZo149atW3h6eqq2bqSiKNy9e7fA8Tt37qj6i3DUqFH5Xr+4uDhVlvzItWzZMmbOnEl4eDh3795l9OjRLFmyxOQ5ueOiHB0dqVWrlvEj97ZanJyc+PHHH1Up/PM6fvw4kPP/VdiHWhYtWkRkZCStW7fmxx9/JDQ0VJVxbpDTdW5vb0/dunX56KOP2Lp1K19++aUqWQBjxozhk08+IS4ujoMHD9KzZ09Gjx6tSlZmZiZly5Y13i5TpowqOyjlOnnyJKNHj8bR0RFHR0dGjx7NyZMnVcsbN24cCxcuxNnZGWdnZ7755hvGjRunSlZaWhqdO3cmKCiItLQ0OnXqROfOnencubOMiXxOpCVSPDEzM7N8Y3ns7OyMuzCoJXddyODgYCwsLFTrJqlRowYjRowAcrbrU5Ner+fzzz9n1KhRxu7569ev8+WXX+Lp6ala7okTJ4iKigLgtddeY8qUKXh7e6uWt3nzZr7//nvatWuHra0tGzZsIDAwkO7du5s8B/IvCaWm3PVLFUVh7dq1+bZ+U2MCQ//+/QFo06YNrq6u+e5T600VaDPEIzk5mXLlypGamoqTkxNXr17Fw8Mj3/aqakhJScn3s9a6dWsWLFigStZ7773HwIED8fb2RqfTER0dreqSYlpNwIqPj8fJyYmqVauyfv160tLSMBgMqo73zN0lR7w4pIgUT6xGjRqsXLmSrKwsTp8+zerVq3F0dFQtr3r16nz66adcuXKFxo0b85///Ic6deqokrVjxw4WLVpUYNcDNdYY7Nq1KykpKXh7e2NpaYmVlRX37t2jc+fOqnbJGgyGfLuC/P3336q+CTAzM8s3waVIkSKqzia+dOkSR48exdvbmzFjxnDy5EnGjh1L7dq1TZrzqGJVjS7Y6OhoMjIymD17trGghJzhCQsXLuSjjz4yeSZoM8QjNDSUhQsXGvdhzi3EFZXX+LSysuLkyZO88847QM4brAd35jGVMWPGsGLFCtatW2ecoNSxY0dVsiBnAlZgYKBxF5mYmBh69Ohh8pxp06Zx4cIFnJycaNKkCa6ursbVJtR09uxZXnvtNcqVK8exY8eIjIzk7bffVnVojng4naJmu7p4qaSlpbFgwQL279+PwWDAxcWFPn36qPbOMysri99++40aNWpQunRpYmJiaN68uSqFSIsWLZg8eXKB3VtMuXNFrtWrV9OxY0fS0tI4deoUxYoVw8HBgSJFijBx4kRGjhxp8kyAqKgowsLCeO+994CcxYFHjhypWhESFhaGTqcjJiaGIUOGsG7dOt58803Vnl+nTp0IDAzExsaGZcuWMWDAAKZOncratWtVyWvfvj3r1q0z3jYYDOj1emNrr6msX7+eI0eOEBMTg7u7u/G4ubk5TZo0oXXr1ibNy3Xu3DnWr1/PsGHDGDBgAPv376dfv3506dJFlbzMzEwsLS3JzMwkIyND1aWhjh49ysCBAyldurRxy7zp06ebtIUwt5X16tWrhd6v1k5RkFNoHTp0CIPBQKNGjVSbgHX//n2OHj1KfHw8hw8f5vr16zg5OdG0aVNVvi8jIiKYPXs2s2bNonTp0rRt25bg4GD++9//8vbbb6syMUo8mhSR4oWVmppKVFQUt27dyjeGSI3WupCQEJYuXap69zzkjOHL7YLN+3lht00ht2iFnC6omzdvYmFhQZ06dVi8eLFqRZ3BYOD777/P96ajQ4cOqq31GRAQwIYNGxg5ciT16tWjXbt2+Pn5sWnTJpPmBAcHc/DgwQLHLSwscHd3Z/bs2SbNyxUbG6vqOqLP0/bt25k/f75xBnNQUJCqM5ghp2i9ePEiBoOBqlWrmnyNz08//ZSFCxfi7u6ebyax2q2s/fr1KzAkJyQkhGXLlqmSl+vMmTMcPnzYuOyVGnu9+/r6smTJEl577TXmzp3LiRMn+Oabb8jIyMDX19fke7qLx5PubPHEwsPDmT9/vnHpDbXGgOUaMGAAJUqUoEaNGqov59CtWzeCg4Np2LCh6mso5i2ItXgPt379emMR+dVXX+UrUg8dOqRarpmZGW3atKF58+aa7NVtbm7Ozp07+fnnnxkwYAC7du1S5U1B7jqXEyZMIDQ01OTnf9CoUaMYP3488+fPL3Tcnlrrbmo5xGP+/PmazGAePnz4I+835RqfCxcuBGDDhg289tprJjvvw/Tt25fTp0+TlJSUb0xpdnY2FSpUMHleUlIS+/btY+/evRw5cgQHBwdcXV2ZPHmyahPaDAaD8bWMi4sztnaquci/eDQpIsUTW758OREREap2w+R148YN4x8WtS1YsICqVauqOmavMFqsdfaoolXNIlbrvazHjRtHeHg4o0ePxs7Ojm3btqm2ODbAiBEjWL16NQcOHCArKwsXFxc6d+5s8sK1ffv2APTu3Vu1VtzCTJo0qdAhHmrQagZzo0aNTH7Ox/H19aVWrVq0bdsWDw+PfAvvm1JYWBi3bt1i4sSJ+d7cWFhYUKZMGZPnubm50bRpU7p06UJYWJhqzysvnU5HRkYGaWlp/Pbbb8ZljFJSUsjOzlY9XxQkRaR4YtWqVcv3i15ttWrV4syZM6pO3smVmZmp2k4jD3qei+Q+mK3mtWzYsIFdu3Zp0goDMHny5HxLCOXdEUgNU6ZM4a+//sLf3x9FUdi0aROXLl0yeetk7sSgKVOmmHyow6O88cYbvPfee5oM8dBqBnPebfL++OMPDh48SFZWFs7Ozqq1nu3evZsDBw6wdetWpk6dirOzM23btjX50AQbGxtsbGxYsGABp06dIi0tDUVRyM7OZu/evSbfQjI0NJR9+/Yxbtw4nJyccHV1xdXVVZWCNVdgYKDxTVXz5s2pXLkysbGxzJgxg8DAQNVyxcNJESmeWHBwMN7e3tSrVy9fi51axde5c+fw9fWlTJkyFClSRNWWLFdXV1auXEmzZs2wtLQ0HlejFebcuXPG7qbExETj54qikJycbPK851W0ar2X9b1797h27Zpx9xG1/frrr0RERBiLrPfff1/VJZPKli3LoUOHqFu3ribdd1oO8dB6BnNERARz587lgw8+wGAw0LdvX3r37q3KXt1mZmY0adKEJk2aEBcXx6RJk+jbty+HDx82eRZotw957vqMmZmZHDlyhH379rFs2TIURaFJkybGLSVNqVOnTtSpU4fk5GTc3NyAnN+hHTp0wM/Pz+R54vGkiBRPbNq0aXh7e6syY7kwuXs+a2Hr1q0AfPfdd8ZjahWsO3fuNPk5H0XrojWX1ntZ37x5E3d3d03edEDOWLOsrCzjc8vOzlZ1OMTx48eNS+GA+mOStRjikTuD+caNG7Rq1YpWrVoZ77tx44ZqXelLly5l/fr12NraAtCrVy+Cg4NVKSJPnTpFVFQUP/74I1WrVqVr1658+OGHJs/JpfU+5JaWltjb21OjRg3u3r1LfHw88fHxquU9uH2rj48P8fHxDBo0iGnTpqmWKwonRaR4YlZWVppuLVexYkXWrFlTYMyZGrRcxFarIjyX1kVrLq33slZjN5xH8fb2Jjg4GC8vLwC2bdtGmzZtVMs7cOBAgWNqbg2oxRCPB9eJhP8bp6vmGwCDwWAsICFn8X21WuxDQ0PR6/WsXbtWk+FAWu1Dvnz5co4cOcKRI0coVaoUjRs3xtXVlYEDB6q64Hiu1NRUNm/ezLp160hOTlblDYB4PCkixRN77733CAsLw83NLV+Xr1r7IU+ePLnAmLPLly+rsiTN+fPn+f777wvMRNVqnKSatC5ac/Xt25e0tDQuXbrEW2+9RXp6OsWKFVMt72GtH2o9/169evH2228TGxuLoij06tWL999/X5UsKHxdSn9/f5OvS5lLiyEeuTOYtd6JpGbNmkycONFYeGzYsEG1sdelS5cmJCRElXMXRqt9yM+dO8eHH37IqFGjVB0H+aCjR4+yZs0afvjhBxwdHbl58ya7d+/WpHAVBck6keKJBQUFFTim0+lUW2Kkbdu2+cacZWVl4e3tzfbt202e1bp1a1q3bl2g4Mg7EF88ndjYWEaPHk12djbr1q2jTZs2TJs2jaZNm6qSl3f5lszMTA4fPkyDBg2Mf0hNpUOHDlStWpVmzZrRtGlT4/Zyanle61LmXdg8l1qtg8eOHePw4cN06tSJXr16cerUKSZPnmwc92Zq6enpzJ49m7i4OBRFwdnZWbWNEzp16sTUqVM1G6t7584d9uzZg5eXFytWrGD//v2EhITg4uKiSb6a9Ho9xYoVo2XLlnh6elKhQgXc3d1lO8TnSRHiBdW6dWvl/v37xtvp6emKl5eXKlnt27dX5byvsoCAACUpKUnR6/WKoijKuXPnFG9vb83yU1JSlC5duqhy7gsXLijLli1TevbsqXTp0kWZP3++cuLECVWyco0fP17V8z8oJiZGs6zAwEDll19+UbZs2aL06tVLSUhIUPz8/DTLV1PLli0VR0dHxdXVVXF3d1datGihuLu7q5bXrVs31c79vPXt21fx8PBQhg0bpvzyyy9KVlaWqq+leDzpzhZPLCEhgdDQUBISEli1ahWDBg3iq6++wt7eXpU8Lcec+fr6MmPGDFxcXPKtxadWV/2rwGAwUK5cOePt6tWra5pfrFgxEhISVDn3m2++yZtvvklwcDD379/nwIEDbN68mQkTJrBmzRpVMt955x0iIiIKHPfx8VElb+rUqcb9l9VmMN1C5MYAACAASURBVBho1qwZgwYNomXLllSsWFGVdf9yd4RydHQsdBcZNSYpaT1WV+tVCrQ0Z84cUlJSiIqKYtq0aQwZMoTMzEyOHz9OnTp1nvflvZKkiBRPbPTo0XTv3p2pU6dStmxZ2rRpwxdffMGqVatMnnX79m3atWtnHHMWFxdHcHCwan8wf/vtN+Mg8VxqdtW/CipUqMDu3bvR6XSkpqayatUqVReuDgoKyjc54/LlyzRv3ly1vLxjMIsVK4anpydt27YlNTVVlS7uvF3aebvr1fqZqFy5MsOHD6devXpYW1sbj6uRV7RoUb777jvi4uIYPXo0y5cvV2Xv7Nx1Ns+cOWPycz9MpUqViIqK4r///S+9evVi586dqv2fgXarFBT2hiYvtZ6jra0twcHBBAcHc/r0aTZu3EiPHj2oVKkSGzduVCVTPJyMiRRPLHcfYh8fH+MvEL1eT2RkpElzTp06Rc+ePfnqq6+MY6KmT5/O5s2bWbx4sSoD4L29vVWboPCq+vvvv5k4cSL79+83jjsLDQ3Fzs7OpDlXr14FcrZByy0i7e3tKV26NMWKFVOtcO3SpQsnTpygcePGKIrCwYMHqVSpEnfu3GHAgAGqztQGuHXrFp9//rlquzo9bItANSabJSYmsn79epo0aUL9+vWZMmUKQUFBJt+u73HLhqmx+sTUqVO5fv06J0+eZP369fTu3Zt33nmHYcOGmTwLeGjru6knmGm5heTjZGZmEhMTQ8uWLTXLFDmkJVI8MWtra65fv278Q33o0CFVFj2eNGkS06ZNw9nZ2Xhs4MCBNGzYkLCwMMLDw02eWaNGDc12x3kVrF69mnLlyjF9+nQCAgK4efMmZ86c4d69eybPyl0eJu/7YZ1OR3JyMpmZmaqto6goClu2bDEWqYmJiYwYMYIVK1YQFBSkehGpZnc9FF4EpKenq5IVGhqKn5+fcXeeIUOGqJKT69ixY1y/fh1PT08sLCz48ccfVZvFv2/fPjZv3oyvry82NjYsXbqUtm3bmryI3L17Ny1atNBslYJHFYlqfZ+MGDHCuNVh7msKOWtVhoeHSxH5HEgRKZ7YsGHD+PTTT7l06RJ6vZ7bt28zc+ZMk+ekpqbmKyBzNWvWjKlTp5o8D3KW+PH19aVcuXJYWlqqvlD1y2zhwoXExsYyZswYIGd5kRUrVrB7924WLlxo/CNgKg/OzLx79y6TJk1i3759jB8/3qRZeSUlJeVr5SxfvjxJSUnY2Niosu+z1t31MTExzJw507h9nsFgID09ndjYWJNn9ejRg4iICKZMmULz5s3x8/NTZYxbbktjhw4dWLduHUWLFgUgJCSE4OBgk+cBxtUlcv/vMjIyVNlK8vjx47Ro0YK4uLhC71ere1nL75O8bwiXL1+eb/UMNd6giseTIlI8sbp167JhwwYuXrxIdnY21apVU6UlMisrC4PBUOAXrcFgIDMz0+R5APPmzVPlvK+iiIgINmzYYBzTZmZmRqVKlejQoQOenp6qZsfGxhIaGoqrqytbtmxRde24+vXrM2jQILy9vTEYDGzbtg0nJyd+/vlnk66Hmdtd7+fnV2h3vVq+/vprxo8fz9KlS+nVqxe7du1S7Q91o0aNaNSoEenp6ezYsYO+fftSokQJAgIC6Nixo8l/z6SkpOSbWJOZmcmtW7dMmpHL09OT//znP9y+fZvw8HC2bNmiSit1//79gZz/t6ysLM6ePYu5uTk1a9ZUdetTLb9P8r45e/CN2vPa3vVVJ0WkeGIPjoHR6XRYW1vj4OBAYGCgyX7RN2zYkLlz5xp/KeaaP3++sbvL1LReqPplZm5unm9SRO/evYGcdQ3VmCwBkJaWRlhYmLH10dXVVZWcvMaOHcuaNWtYt24d5ubmNG7cmPbt2/Prr78yefJkk+U8r+76EiVK4OLiwpEjR/jnn38YMmQIrVu3ViULcsa0RkZG8uuvv+Lm5kbr1q3Zv38/vXv3NvkM58DAQPz9/XFzc0NRFHbv3q1aS2TPnj3Zu3cvFStW5Nq1a/Tr10/VWe/79+9n6NCh2NnZYTAYSE1NZebMmQW2CzQVLb9P8haKUjS+GKSIFE/M3Nyc27dvG7tFoqOjuXv3LmZmZowZM8ZkA6kHDhxIz549iYiIwNHRkSJFinDq1Clee+01FixYYJKMB+XtAtJi5uvLzGAwcOfOHWMrYO44pX/++UeVbry8rY9RUVGqFaoPsrCwwNfXlw8++MBY4CUlJZm8i/l5dddbW1tz4cIFHBwcOHjwIC4uLqr1BLRo0QJ7e3v8/f0ZPXq0cTa4s7Mz/v7+Js/75JNPcHFx4eDBg+h0OmbNmqXKeOjs7Gyys7Np1qwZTk5O/Prrr7z55psmz8nrq6++4ttvvzU+n+PHjzNmzBg2bdqkSp6W3yeZmZlcu3bN2Ct17do148+eWpni0WR2tnhi/v7++ZZQUBSFwMBANmzYQNu2bdmyZYvJshRF4cCBA5w+fRozMzNq165NgwYNTHb+x1F75uvLbMGCBZw4cYJJkyb9v/buPSrmff0D+HsqFWUrMW33S7EGoTLu98reqUM394rFppMIXcR2K5JLRZI49jmIdNqOy6Zcdk4nbNfSlt1FwgnJpvopQzczNd/fH9bMatyaTvOdb+V5rbXXrm81z5PFzDOf7+fzPPJCsqKiAqtXr4alpSXmz5+v0ngCgQBaWlrg8/mf7P3H1r7Wv/3tb/jpp59gYGAgXylkex9t3YI5ICCA1dv1aWlpiIuLQ1hYGGbPno2CggK4uLiwcqq4oKAA3bt3V/njfo5YLMaVK1dQUVEB4H2xV1hYiOXLl6ssRlZWFry8vLB161aYm5vD0dERHTt2RGlpKVauXAkbGxuVxapL1kWjvmuqos6/J1ZWVh+tysvQHnZu0EokUVplZSVKSkrkDaRfvXqFd+/eAYDKGwPzeDyMHDkSI0eOVOnjKovtk68tmYeHB4KCgjB27FiYmJiAx+Ph0aNHcHBwUHkBCYCzF44TJ04gOTkZ7du3Zz0WF7frZfsUAeDkyZMQiURo164dK7Fev36NnTt3oqysTKFAYKtPq6+vL0QiEQoKCiAUCpGamgpLS0uVxggNDUVkZCQsLS0RGxuLdu3aIT4+HiUlJfjrX//KWhEpFAqxdu1azJgxA5qamjh37hy6dOki37Kj6gEK6vx7QuMNmx4qIonSvL294ezsDAsLC0ilUmRnZ2Pt2rWIiorCqFGjuE6vUT48+VpYWMjqydeWTFNTE8HBwVi6dCkyMzMBAGZmZqxN0OBq32qnTp1Ye7Gsi6vb9ZmZmTh48KBaCrtVq1bBzc0NpqamatnrlpeXh4sXLyIkJAQuLi5YsWIFVqxYodIYIpFIXpjevHlTvq2jY8eOrN56le2R/bCTxe7du1U6QGH9+vUIDg5WeO6si603ADdv3gSfz4eJiQkAIDY2FqamppwtOHztqIgkSrOzs8OIESPw+++/Q0NDA5s2bUL79u0xdOhQGBgYcJ1eo3h7e8s/5vF4MDQ0VPuYvpbG2NgYkyZN4joN1vTs2RNz5szB8OHDFQ6Vqbph9fz586GlpYVr167h+vXr8uts3z5XZ2Gnq6sLV1dXVmPUZWRkBB6Ph169eiEvLw+Ojo4Qi8UqjVF3r97t27flB8wkEon8NjobYmNjFXKoqKhgZdvDzJkzASg+d7Lt/PnziIiIQEREhPyakZER1q9fj5UrV1KfSA5QEUmUVlpaioSEBFRUVIBhGOTk5KCwsFClJ1G5Ym5ujvz8fAgEAiQmJiIlJQWLFi1Sy61K0jwZGxvD2NiY9Thc3a5XR2Ena1/Ur18/xMTEwNraGpqamvKvszVtqE+fPggODsbs2bPh7++P4uJilRfKQ4cOxcaNGyGRSGBsbIyBAweiqKgI+/btw5gxY1Qaq65Lly4hPT0dXl5e8kb/q1atgrOzs0rjyDpl9OjRA0eOHMHKlSvx7NkzREVFISAgQKWxZP7xj3/g6NGjCv/u7OzsMGjQICxbtoyKSA7QwRqitLlz56JTp064e/cubGxscPnyZQwcOBDbtm3jOrVGW758Obp27Yrvv/8e/v7+cHBwQGZmJvbv3891aoSolayw27NnD/r27ctqYWdlZfXZr7G1ypqfnw89PT08e/YMQqEQKSkpuHz5Mp4+fYrDhw+rLI5YLMbhw4fxf//3f5g7dy66dOmCiIgIFBUVYcOGDaz1+HRxcUFISAiysrKQnp6ODRs2wN3dnbWDNe7u7rC3t8esWbMgFotx+vRp/Prrrzh48KDKYzk5Ocnnn3+o7jheoj60EkmUVlxcjCNHjmD79u347rvvsHDhQsybN4/rtFSisLAQkZGRCAsLw7Rp0+Dh4cFKaxHS/MleyAQCwSdPg7PVt1FdZH0pgfetr+rubVN1YafugxJRUVHy4kY2Rzs/Px9nz56FhYWFSmNpa2tj0aJFCtd8fHxUGuNzBAIBoqKiMHXqVOjp6bG6B1MkEmHWrFkA3v/OM2bMQHx8PCuxZLfnP9wXXF5eTi1+OEJFJFGa7BBBr169cP/+fQwePJjjjFSntrYWpaWlSE5ORlRUFEpKSuQnzwmpS7YScv/+fY4zYUd8fDxCQ0Px4MEDWFhYwN/fH9988w0rsYqKihAaGoqHDx/CwsICfn5+rMUC3k9TSkpKQnFxMXbv3o1Dhw6hqKgIkZGRGDt2LGtx1alDhw4IDg5GdnY2wsLCsG3bNta2BQDvtz1cuXJFfhDxxo0b8nGSqubg4AAfHx+sX78e3bp1AwC8fPkSQUFBmDx5MisxyZepvvMvabFGjBiBZcuWYfTo0Th48KBCU+Dm7ocffsCMGTMwfvx49O3bF25ubvDy8uI6LdKEFRQUICEhAQzDYMOGDXBxcUF2djbXaTXamjVrwOfz4efnB4lEorIhAl+K5evrC7FYzGosANDT0wOfz4eZmRkyMzNhamqK06dPt5gCEgB27NiBgQMHIjY2Fm3atEG3bt2wY8cO1uJt3LgRYWFhGD58OIYPH47Q0FAEBQWxEmv+/PkQCASYMmUKhg4ditGjR8POzg59+/al52uO0J5I0iCypsA5OTm4ffs27OzswOfzuU5L5WpraxX2gRHyIVdXV0yfPh36+vo4fPgwli9fjvDwcPz8889cp9Yof/nLX3D27FkA708SOzo64ty5c80+FqC4b27y5Mm4cOECa7E+VF5eDi0tLbW88U5ISMB///tfeHp6IikpSS2Tt/7880/weDzWWnnVVVlZifz8fGhoaMDExAQ6OjqsxySfRiuRRGk1NTXIz8/H6dOn8fDhQxgYGODGjRtcp8UKKiBJfd69ewdHR0dcunQJU6ZMgVAoVHmbGC60atVK4eO6nzfnWIDivGV13UXJy8uDk5MTrK2tMX78ePlUF7aEh4fjt99+w8WLF1FbW4uTJ0+yevixoKAA06ZNg5OTExwcHODo6IjHjx+zEuuf//wngPfDIHR0dNC/f395ARkSEsJKTPJlVEQSpfn5+SE6Ohq3bt1Camqq/D9CvkaamppISkrC5cuXMWHCBCQnJ7MyG5xr6mj+ra5YDx8+hLW1NaytrRU+trKygrW1NSsxAwMDsWLFCvnz5YIFC7BmzRpWYgHAtWvXEBYWBh0dHejr6+PQoUP47bffWIsXGBiIhQsXIjU1FWlpafDw8MCGDRtYiXX8+HH5xx+2EUpPT2clJvkyOlhDlJaXl4cLFy6o9UVF3dgc2UValk2bNiEmJgYbNmwAn8/HuXPnsHnzZq7TajRZcSVTVFQEa2trVpqbqzMWACQlJan08ZTx7t07helXkyZNQnR0NGvxZG9kZM/TYrGY1Tc3ZWVlsLW1lX9uZ2eHffv2sRKr7u472onXNFARSZRmYmKCkpKSFrkHMjc3Fz4+PqiursaxY8fg5uaGXbt2YcCAAVynRpqo0NBQHDhwQP553SkazZk6Cy11F3XqHJEp67cpEAjw008/Ydq0adDU1ERiYiKEQiFrcW1tbbFixQqIRCLExMQgISEB9vb2rMXT1tZGTk6O/LkyOzubtdPZdbXkxYzmhIpIorTq6mrY2tqib9++CmPe2JqRqk6bN29GdHQ0/Pz8YGxsjKCgIAQGBuLEiRNcp0aaqKqqKrx48UItBwnUSZ2FFldzz9VB1m+TYRikpqYqHLji8XhYt24dK3E9PDxw9epVdO7cGS9evIC3tzcmTpzISizg/Ql7b29vGBgYgGEYiEQi1t5QUeHY9NDpbKK0tLS0T14fNmyYmjNRPWdnZ5w6dUrh9ObUqVORkJDAcWakqbK1tcXTp09hZGQEHR0d1mdZE1If2TSeumMBX716hcjISGzatIm1uBKJBE+ePIFUKkWvXr0UFhlUyczMDMbGxmAYBsXFxfLfk2EYlJSUICsri5W45PNoJZIobdiwYbh37x4qKyvBMAxqa2tRWFjYIopIAwMD3L9/X/5ONyEhgfZGki+qeyubkM958uQJjh49Kn/elEqlKCwsRFxcnErj1J3GEx0djVGjRuHAgQOIjo5W+TQemUuXLsHU1BTdunXD06dPceLECfTv3x+LFy9m5aS9bPtDeXk5rl69itatW2PcuHEt8kBbc0ErkURp69atQ1paGkQiEXr37o379+/D0tKyRbyYFhQUYNWqVcjKyoKuri569OiBsLAw9O7dm+vUSBP1uTm96ujJR5oPZ2dnTJgwAZcuXYKTkxP+/e9/w8TEROUNua2trREfHy+fxiOVSlFUVISAgABWmqkfOHAA58+fx/bt21FTU4NZs2Zh7dq1yM3NhaamJtauXavymK9evcKyZcvw6NEjdO/eHTweD0+ePIG5uTl27NiBtm3bqjwm+TJaiSRKu3HjBpKSkhAcHIy5c+eiqqqK1f5j6tS9e3fEx8ejsrISUqkU+vr6XKdEmri67a0kEgl+//13CIVCKiKJAolEgmXLlqGmpgb9+/fHjBkz4OLiovI4smk8fD4fmZmZcHR0xP79+1nreXvmzBkcO3YMrVu3Rnh4OKysrDB9+nQwDAM7OztWYu7YsQNDhgxBTEyMfKVTLBYjKioKISEhLeb1qDmhIpIojc/no1WrVjAxMUFeXh7s7e3x9u1brtNSCXd3d4VN2zweD7q6uujduzc8PT3p1jb5yIcj+l6/fg0fHx+OsiFNVevWrSEWi9GzZ0/k5OSwdjK77i1dQ0NDrF69mpU4MjweT34KOzU1FXPmzJFfZ0tGRsZHU4a0tbXh6+sLBwcH1uKSz6MikijN2NgY+/fvx8iRIxEWFgYALWJCBwCYmppCS0tLvkJw9uxZvHz5EsbGxli7di327NnDcYakqWvTpg2eP3/OdRqkiZk6dSo8PT0RHh6OmTNn4urVqwoHX1RF3dN4NDU18ebNG1RWViI3NxejR48GADx//hxaWuyUFp8bb8jj8WhfJEeoiCRKCwkJwZUrVzBo0CB89913OHv2rMr39XDljz/+wKlTp+SfCwQCuLi4IDw8/LN738jXre7qNcMwePbsmUJTaUKA961+HB0doa+vj9jYWGRlZWHMmDEqj1O3cbusaTsA1roGeHh4wNHRETU1NZg2bRr4fD7Onz+PiIgILFmyRKWxZL60ykntf7hBRSRRmr6+vrxprbu7O9zd3TnOSHUkEgkePnyIPn36AHj/hCyVSlFdXQ2JRMJxdqQpkTWRdnZ2lr9wde3aFQYGBmjTpg2XqZEmSCQS4dy5cygrK5NPWcnLy8PSpUtVGkfdjdttbW1hYWGBsrIyCAQCAO/3ZW7evBnDhw9nJeaHE45kZC1+iPrR6WxSL4FAoPAuT/bOVvb/3NxcDrNTjbS0NAQEBMDIyAhSqRRv3rxBaGgoUlJS0K5dO3h4eHCdImkirKys5H//ZXg8HkpKSiCRSFrEvweiOm5ubmjfvj369Omj8Dyq6iLya1DfdpGW3Ly+qaKVSFIvd3d3pKenw9zcHHZ2dhAKhS3u1kFlZSWSk5Px4MEDaGhowMTEBK1atYKlpWWL+11J46SkpCh8XlFRge3bt+PatWsIDg7mKCvSVIlEIhw9epTrNFoEKhKbHlqJJEpLT0/H+fPncffuXQiFQtjb22Pw4MFcp6US9vb2OHfuHNdpkGbm5s2bWLduHUaPHo2AgABqDUU+4ufnh/nz58PMzIzrVAhROSoiSYNJpVKkpqYiMjISxcXFH63MNEeenp4wNDTE4MGDFU42Us8/8imVlZXYtm2bfPVRdjKVEBnZtofq6mqUlpbC2NgYmpqaNB6TtCh0O5s0SHZ2NpKSkpCcnIzOnTvDy8uL65RUwtDQEMD7U9p1URFJPlR39TExMRF6enpcp0SaoNjYWK5TIIR1tBJJ6vXHH3/g119/RUpKCrp27YrJkyfDxsYGBgYGXKfGqurqarX0WyPNi0AggJaWFvh8/icPnNEKE/lQYmIiHj16BE9PTyQlJdGbU9JiUBFJ6iUQCNCpUydYWVnJV+zqagmnDFNSUrBr1y5UVlaCYRh5e5+bN29ynRppYuiEKGmI8PBwvHz5Ejk5OTh+/DgWL16MAQMGsD5RhhB1oNvZpF5Llixp8SeUt27diuDgYBw6dAienp5ITk5GVVUV12mRJoiKRNIQ165dwy+//AInJyfo6+vj0KFDmDp1KhWRpEWgIpLUy9vbm+sUWNe2bVuMGDECd+7cwdu3b7Fy5UrY2dlxnRYhpJmTjeOTvREXi8U0oo+0GPQ3mRC8nzX7+PFjmJiYIC0tDWKxmCbVEEIazdbWFitWrIBIJEJMTAzc3Nzkk78Iae5oTyQheD+xJi4uDmFhYZg9ezYKCgrg4uJCt5wIIY129epV3LhxA1KpFCNGjMDEiRO5TokQlaAikjSYSCRCu3btuE6DVV/D70gIYVd+fj709PRgbGwsv/bq1StERkZi06ZNHGZGiGrQnkiitNzcXPj4+KC6uhrHjh2Dm5sbdu3ahQEDBnCdWqNlZmbi4MGDKCsrU5iJfOTIEQ6zIoQ0V1FRUTh48CAAIDo6GqNGjcKBAwcQHR0NCwsLjrMjRDVoJZIozdXVFZs2bYKfnx9Onz6N69evIyIiAidOnOA6tUabPHky3NzcYGpqqnASfdiwYRxmRQhprqytrREfH4/i4mLs3r0bUqkURUVFCAgIwNixY7lOjxCVoJVIorSqqiqYmJjIPx89ejS2b9/OYUaqo6urC1dXV67TIIS0EHp6euDz+eDz+cjMzISjoyP2798PTU1NrlMjRGWoiCRKMzAwwP379+UrdQkJCc1+3+Cff/4JAOjXrx9iYmJgbW2t8CTfuXNnrlIjhDRjddv4GBoa0iE90iJREUmUFhQUhFWrVuHhw4cYMmQIevbsibCwMK7TahQ3NzfweDwwDINbt24p7IGkEXaEkP9V3W0xND6VtFS0J5I0WGVlJaRSKQBAX1+f42xURyKRoFWrVpBIJBCLxdDT0+M6JUJIM2VmZiY/lV1UVCT/mGask5aEikiitEuXLiE9PR1eXl6YNm0aSktLsWrVKjg7O3OdWqNduHABe/fuRWJiIgoKCuDu7o7169fDxsaG69QIIc0QzVgnXwMqIonSXFxcEBISgqysLKSnp2PDhg1wd3fHqVOnuE6t0aZMmYJDhw6hQ4cOAN73cluwYAHOnDnDcWaEEEJI00RjD0mDCAQCXL58GVZWVtDT02sxowElEom8gAQAIyMj0PsrQggh5PPoYA1RWocOHRAcHIzs7GyEhYVh27ZtLeb08pAhQ+Dr64spU6aAx+Ph3LlzMDc35zotQgghpMmi29lEaeXl5UhOToaFhQV69OiBuLg4ODo6togDKGKxGLGxsbh9+za0tLQwdOhQzJ49G9ra2lynRgghhDRJVEQSpYnFYly5cgUVFRUAgNraWhQWFmL58uUcZ6YaxcXF4PP5uH37Nh48eAAXFxdqzUEIIYR8Bt3OJkrz9fWFSCRCQUEBhEIhUlNTYWlpyXVaKhEYGAiJRIIFCxZg5cqVGDVqFDIyMhAeHs51aoQQQkiTRAdriNLy8vJw5MgRTJo0CQsXLkR8fHy9bSyai6ysLISEhODChQtwcXHBli1b8PjxY67TIoQQQposKiKJ0oyMjMDj8dCrVy/k5eWhW7duLeZ0dm1tLaRSKf7zn/9g3LhxqKqqQlVVFddpEUIIIU0W3c4mSuvTpw+Cg4Mxe/Zs+Pv7o7i4uMW0wXF0dMSYMWNgaWmJwYMHw87ODjNnzuQ6LUIIIaTJooM1RGm1tbXIyMiAUChESkoKbty4gRkzZqBv375cp6YSUqkUGhrvF+dLS0vRvn17jjMihBBCmi4qIolSRCIRamtr5YVVWloaTE1Nm32htX79egQHB8Pd3R08Hu+jrx85coSDrAghhJCmj25nk3rdu3cPHh4e2LJlC8aNGwcAuH79Ovz8/PD3v/8dAoGA4wz/d7Jb1t7e3hxnQgghhDQvtBJJ6jVv3jx4eXlh+PDhCtevXr2KAwcOICYmhpvEVOjBgwfIz8+Hrq4uTExM0K1bN65TIoQQQpo0Wokk9Xrz5s1HBSQAjB07ttn3UXz16hWWLVuGhw8fokePHuDxeHj8+DHMzc2xc+dOtG3blusUCSGEkCaJWvyQetXU1EAqlX50XSqVNvsWPzt27MCQIUNw/fp1HD9+HP/6179w/fp1CAQChISEcJ0eIYQQ0mRREUnqNXToUOzZs+ej63v37oWZmRkHGalORkYGfH190apVK/k1bW1t+Pr64t69exxmRgghhDRtdDub1MvX1xceHh44ffo0BAIBdHR0cO/ePbRv3x779u3jOr1G0dHR+eR1Ho8nb/dDCCGEkI9REUnqpa+vj7i4ONy6dQu5ubnQ0NCAq6srhEIh16k12qfa+ijzNUIIIeRrR6ezyVfNzMwMxsbGH11nGAYlJSXIysriICtCCCGk6aMiknzVnj9//sWvd+nSRU2ZEEIIIc0LFZGEEEIIIaTBf8c99QAABU9JREFU6OQAIYQQQghpMCoiCSGEEEJIg9HpbELIV6u2thZHjhxBYmIiamtrIZFIMHHiRCxfvhza2toAgJSUFCxevBgRERGws7MD8H6Kk7u7OwCgsrISRUVF6NWrFwBg1KhRcHV1xaRJk9C3b9+PYh4/fhza2tqoqanB4cOHkZCQgJqaGlRVVUEgEMDf3x+9e/eWf//Nmzexd+9eFBUVQVdXF0ZGRliyZIm8O8KpU6cQEhKCrl27Anh/KKy8vBxCoRDBwcHIzc3FvHnzEB8fj/79+8sfNy4uDrGxsTh58iT09PRY+NMlhLR4DCGEfKXWrVvHeHt7M2/evGEYhmEqKiqYxYsXM/7+/vLv+eGHHxg/Pz9m+vTpn3yMW7duMfb29grXnj17xpibm38xtq+vL+Pl5cWUlpbKryUkJDBjx45lysvLGYZhmOTkZMbGxoa5c+eO/HsyMjKYCRMmMJcvX2YYhmFOnjzJeHh4KDx2dXU14+TkxMTHxzMMwzB79uxh7O3tmXfv3jEMwzCPHz9mhg0bxuTm5n4xR0II+RK6nU0I+SoVFhYiMTERW7Zskc9Ib9OmDTZu3AgbGxsAwLNnz5CWloYff/wRT58+xd27d1USOzs7Gzdv3kR4eDgMDQ3l16dMmQIPDw+Ul5cDAEJDQ7Fu3TpYWFjIv8fc3Bxr1qxBaGjoZx//9evXKC8vR7t27QAAixcvhoGBASIiIlBbW4uAgAD4+vpCIBCo5PchhHyd6HY2IeSrlJOTA1NTU+jr6ytc79ixI77//nsAQHx8PCZMmAAjIyPY2dkhJiYGu3btUurxq6ur4eDgoHDN0tISgYGBSE9Px5AhQ9C6deuPfs7NzQ0AUFZWhidPnmDo0KEffc/IkSOxdOlSiEQiAEB6ejocHBzw7t07vH79Gj179sSCBQswefJkAICGhgbCwsLg5OSE0tJSdO3aFTNnzlTq9yCEkM+hIpIQ8lXS0NCAVCr97NfFYjFOnTqFLVu2AACcnJwwe/ZsvHjxAp06dar38XV1dXHmzJnPfr3uRKT8/Hz4+PgAAN6+fYuFCxfKC8CampqPflYikSg8hlAoxP79+yGVSrF3716cPXsWtra2Cj/TqVMn+Pv7Y+fOnbh48WK9+RNCSH3odjYh5Ks0aNAg5Ofny28dyxQVFcHDwwPnz5/HmzdvEBwcDCsrK6xYsQI8Hg+xsbGNjm1hYYE7d+7Ii8HevXvjzJkzOHPmDIYNG4aqqioYGhqiV69eSEtL++jnb926BRMTE3zzzTcK1zU0NLB06VJ06dIFq1ev/ujnunXrhvbt23+0+koIIf8LKiIJIV8lY2NjTJkyBWvWrJEXkuXl5QgKCoKBgQF+/vlneHp64tKlS0hJSUFKSgqCgoJw/PhxVFZWNir24MGDMXz4cAQEBKC0tFR+/dGjR8jNzYWmpiYA4Mcff8SWLVsU9mJmZGRg27Zt8Pf3/+zjBwYG4vr160hOTm5UnoQQ8iV0O5sQ8tUKDAzE3r17MWvWLGhqakIsFsPGxgb29vaYOXMm9u7dq/D9jo6O2LdvH3755Re4urp+8bE/tScSALZt24Z+/fohLCwMcXFx8PDwQE1NDUQiEb799lvMmTMHzs7OAIDx48dj+/btiIyMRFFREaRSKb799lts374dI0aM+Gzs7t27Y9GiRdi6dSvGjh0LHR2d/+FPhxBCvozGHhJCCCGEkAaj29mEEEIIIaTBqIgkhBBCCCENRkUkIYQQQghpMCoiCSGEEEJIg1ERSQghhBBCGoyKSEIIIYQQ0mBURBJCCCGEkAajIpIQQgghhDTY/wM016bFwyHNowAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 720x432 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "sns.heatmap(has_note_type.corr())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Nothing too severe, let's see last discharge notes for some alive and deceased patients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "deceased_notes = notes[(notes.CATEGORY == 'Discharge summary') & notes.SUBJECT_ID.isin(DECEASED_TO_DATE)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols_to_use = ['SUBJECT_ID', 'DATE']\n",
    "for _, row in deceased_notes.sort_values(cols_to_use, ascending=[True, False])\\\n",
    "        .drop_duplicates(cols_to_use).iloc[:3].iterrows():\n",
    "    pass\n",
    "#     print(row['TEXT'])\n",
    "#     print('\\n' * 3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "If we print the previous results we see that deceased patients don't have some explicit mention of their death later, so there should not be major data leaks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False    23409\n",
       "True      1618\n",
       "Name: TEXT, dtype: int64"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "deceased_notes.TEXT.map(lambda x: 'deceased' in x.lower()).value_counts()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "That helps us to conclude that direct pointers to death can rarely be found in discharge notes"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Splitting discharge notes into parts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols_to_use = ['SUBJECT_ID', 'DATE']\n",
    "discharge_notes = notes[notes.CATEGORY == 'Discharge summary']\\\n",
    "    .sort_values(cols_to_use, ascending=[True, False])\\\n",
    "    .drop_duplicates(cols_to_use)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(41127, 55746)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "discharge_notes.SUBJECT_ID.nunique(), discharge_notes.shape[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "52650"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "discharge_notes.TEXT.map(lambda x: 'history ' in x.lower()).sum()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's try to use only the history part of discharge notes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "def prepare_text_for_tokenizer(text: str):\n",
    "    cleaned = ('. ').join(text.splitlines())\n",
    "    removed_symbols = re.sub('[\\[\\]\\*\\_#:?!]+', ' ', cleaned)\n",
    "    removed_spaces = re.sub(' +', ' ', removed_symbols)\n",
    "    removed_dots = re.sub('\\. \\.| \\.', '.', removed_spaces)\n",
    "    removed_duplicated_dots = re.sub('\\.+', '.', removed_dots)\n",
    "    return removed_duplicated_dots"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "discharge_notes['TO_TOK'] = discharge_notes.TEXT.map(prepare_text_for_tokenizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "discharge_notes['HISTORY'] = discharge_notes['TO_TOK'].map(lambda x: x.lower().split('history')[-1])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Feature construction"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We need to take into account only the events that happened during the observation window. The end of observation window is N days before death for deceased patients and date of last event for alive patients. We can have several sets of events (e.g. labs, diags, meds), so we need to choose the latest date out of those."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "OBSERVATION_WINDOW = 2000\n",
    "PREDICTION_WINDOW = 50"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "def define_train_period(*feature_sets, \n",
    "                        obs_w: int = OBSERVATION_WINDOW, \n",
    "                        pred_w: int = PREDICTION_WINDOW) -> (dict, dict):\n",
    "    \"\"\"Create SUBJECT_ID -> earliest_date and SUBJECT_ID -> last date dicts.\"\"\"\n",
    "    cols = ['SUBJECT_ID', 'DATE']\n",
    "    all_feats = pd.concat([feats[cols] for feats in feature_sets])\n",
    "    last_date_base = all_feats.groupby('SUBJECT_ID').DATE.max().to_dict()\n",
    "    last_date = {subj_id: date\n",
    "                 for subj_id, date in last_date_base.items()\n",
    "                 if subj_id not in DECEASED_TO_DATE}\n",
    "    subtracted_pred_w = {subj_id: date - dt.timedelta(days=pred_w)\n",
    "                         for subj_id, date in DECEASED_TO_DATE.items()}\n",
    "    last_date.update(subtracted_pred_w)\n",
    "    earliest_date = {subj_id: date - dt.timedelta(days=obs_w)\n",
    "                     for subj_id, date in last_date.items()}\n",
    "    return earliest_date, last_date"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "def clean_up_feature_sets(*feature_sets, earliest_date: dict, last_date: dict) -> list:\n",
    "    \"\"\"Leave only features from inside the observation window.\"\"\"\n",
    "    results = []\n",
    "    for feats in feature_sets:\n",
    "        results.append(feats[(feats.DATE < feats.SUBJECT_ID.map(last_date))\n",
    "                             & (feats.DATE >= feats.SUBJECT_ID.map(earliest_date))])\n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "use_feature_sets = [diag_preprocessed, lab_preprocessed, meds_preprocessed]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "earliest_date, last_date = define_train_period(*use_feature_sets)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "diag, lab, med = clean_up_feature_sets(*use_feature_sets, earliest_date=earliest_date, last_date=last_date)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Feat calculations"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We are going to do a train test split based on patients to validate our model. We will only use those features that appear in the train set. Also, we will only use features that are shared between many patients (we will define \"many\" manually for each of the feature sets).  \n",
    "\n",
    "This way we will lose some patients who don't have \"popular\" features, but that's fine since our goal is to compare similar patients, not to train the best model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "TRAIN_SIZE = 0.8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "TRAIN_IDS, TEST_IDS = train_test_split(list(SAMPLE_IDS), train_size=TRAIN_SIZE, random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_feats(df: pd.DataFrame, agg: list, train_ids: list = None, low_thresh: int = None) -> pd.DataFrame:\n",
    "    \"\"\"Build feature aggregations for patient.\n",
    "    \n",
    "    Args:\n",
    "        agg: list of aggregations to use\n",
    "        train_ids: if not empty, only features that exist in the train set \n",
    "            will be used\n",
    "        low_thresh: if not empty, only features that more than low_thresh\n",
    "            patients have will be used\n",
    "    \"\"\"\n",
    "    cols_to_use = ['SUBJECT_ID', 'FEATURE_NAME']\n",
    "    print(f\"Total feats: {df.FEATURE_NAME.nunique()}\")\n",
    "    \n",
    "    if train_ids is not None:\n",
    "        train_df = df[df.SUBJECT_ID.isin(train_ids)]\n",
    "        train_feats = set(train_df.FEATURE_NAME)\n",
    "        df = df[df.FEATURE_NAME.isin(train_feats)]\n",
    "        print(f\"Feats after leaving only train: {len(train_feats)}\")\n",
    "        \n",
    "    if low_thresh is not None:\n",
    "        deduplicated = df.drop_duplicates(cols_to_use)\n",
    "        count = Counter(deduplicated.FEATURE_NAME)\n",
    "        features_to_leave = set(feat for feat, cnt in count.items() if cnt > low_thresh)\n",
    "        df = df[df.FEATURE_NAME.isin(features_to_leave)]\n",
    "        print(f\"Feats after removing rare: {len(features_to_leave)}\")\n",
    "    \n",
    "    grouped = df.groupby(cols_to_use).VALUE.agg(agg)\n",
    "    return grouped"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "def pivot_aggregation(df: pd.DataFrame, fill_value = None, use_sparse: bool = True) -> pd.DataFrame:\n",
    "    \"\"\"Make sparse pivoted table with SUBJECT_ID as index.\"\"\"\n",
    "    pivoted = df.unstack()\n",
    "    if fill_value is not None:\n",
    "        pivoted = pivoted.fillna(fill_value)\n",
    "    \n",
    "    if use_sparse:\n",
    "        pivoted = pivoted.astype(pd.SparseDtype(\"float\", fill_value))\n",
    "\n",
    "    pivoted.columns = [f\"{col[-1]}_{col[0]}\" for col in pivoted.columns]\n",
    "    return pivoted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total feats: 6734\n",
      "Feats after leaving only train: 6399\n",
      "Feats after removing rare: 1489\n"
     ]
    }
   ],
   "source": [
    "diag_built = build_feats(diag, agg=['count'], train_ids=TRAIN_IDS, low_thresh=30)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "diag_final = pivot_aggregation(diag_built, fill_value=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ICD9_00845_count</th>\n",
       "      <th>ICD9_0088_count</th>\n",
       "      <th>ICD9_0380_count</th>\n",
       "      <th>ICD9_03811_count</th>\n",
       "      <th>ICD9_03812_count</th>\n",
       "      <th>ICD9_03819_count</th>\n",
       "      <th>ICD9_0382_count</th>\n",
       "      <th>ICD9_0383_count</th>\n",
       "      <th>ICD9_03840_count</th>\n",
       "      <th>ICD9_03842_count</th>\n",
       "      <th>ICD9_03843_count</th>\n",
       "      <th>ICD9_03849_count</th>\n",
       "      <th>ICD9_0388_count</th>\n",
       "      <th>ICD9_0389_count</th>\n",
       "      <th>ICD9_04101_count</th>\n",
       "      <th>ICD9_04102_count</th>\n",
       "      <th>ICD9_04104_count</th>\n",
       "      <th>ICD9_04109_count</th>\n",
       "      <th>ICD9_04110_count</th>\n",
       "      <th>ICD9_04111_count</th>\n",
       "      <th>ICD9_04112_count</th>\n",
       "      <th>ICD9_04119_count</th>\n",
       "      <th>ICD9_0413_count</th>\n",
       "      <th>ICD9_0414_count</th>\n",
       "      <th>ICD9_04149_count</th>\n",
       "      <th>ICD9_0415_count</th>\n",
       "      <th>ICD9_0416_count</th>\n",
       "      <th>ICD9_0417_count</th>\n",
       "      <th>ICD9_04184_count</th>\n",
       "      <th>ICD9_04185_count</th>\n",
       "      <th>ICD9_04186_count</th>\n",
       "      <th>ICD9_04189_count</th>\n",
       "      <th>ICD9_042_count</th>\n",
       "      <th>ICD9_0539_count</th>\n",
       "      <th>ICD9_0549_count</th>\n",
       "      <th>ICD9_07030_count</th>\n",
       "      <th>ICD9_07032_count</th>\n",
       "      <th>ICD9_07044_count</th>\n",
       "      <th>ICD9_07051_count</th>\n",
       "      <th>ICD9_07054_count</th>\n",
       "      <th>ICD9_07070_count</th>\n",
       "      <th>ICD9_07071_count</th>\n",
       "      <th>ICD9_0785_count</th>\n",
       "      <th>ICD9_07999_count</th>\n",
       "      <th>ICD9_1101_count</th>\n",
       "      <th>ICD9_1103_count</th>\n",
       "      <th>ICD9_1120_count</th>\n",
       "      <th>ICD9_1121_count</th>\n",
       "      <th>ICD9_1122_count</th>\n",
       "      <th>ICD9_1123_count</th>\n",
       "      <th>ICD9_1125_count</th>\n",
       "      <th>ICD9_11284_count</th>\n",
       "      <th>ICD9_11289_count</th>\n",
       "      <th>ICD9_1173_count</th>\n",
       "      <th>ICD9_1179_count</th>\n",
       "      <th>ICD9_135_count</th>\n",
       "      <th>ICD9_1363_count</th>\n",
       "      <th>ICD9_138_count</th>\n",
       "      <th>ICD9_1505_count</th>\n",
       "      <th>ICD9_1508_count</th>\n",
       "      <th>ICD9_1510_count</th>\n",
       "      <th>ICD9_1541_count</th>\n",
       "      <th>ICD9_1550_count</th>\n",
       "      <th>ICD9_1551_count</th>\n",
       "      <th>ICD9_1552_count</th>\n",
       "      <th>ICD9_1570_count</th>\n",
       "      <th>ICD9_1578_count</th>\n",
       "      <th>ICD9_1622_count</th>\n",
       "      <th>ICD9_1623_count</th>\n",
       "      <th>ICD9_1625_count</th>\n",
       "      <th>ICD9_1628_count</th>\n",
       "      <th>ICD9_1629_count</th>\n",
       "      <th>ICD9_1748_count</th>\n",
       "      <th>ICD9_1749_count</th>\n",
       "      <th>ICD9_1830_count</th>\n",
       "      <th>...</th>\n",
       "      <th>ICD9_V3101_count</th>\n",
       "      <th>ICD9_V3401_count</th>\n",
       "      <th>ICD9_V420_count</th>\n",
       "      <th>ICD9_V422_count</th>\n",
       "      <th>ICD9_V427_count</th>\n",
       "      <th>ICD9_V4281_count</th>\n",
       "      <th>ICD9_V4282_count</th>\n",
       "      <th>ICD9_V4283_count</th>\n",
       "      <th>ICD9_V433_count</th>\n",
       "      <th>ICD9_V4364_count</th>\n",
       "      <th>ICD9_V4365_count</th>\n",
       "      <th>ICD9_V440_count</th>\n",
       "      <th>ICD9_V441_count</th>\n",
       "      <th>ICD9_V442_count</th>\n",
       "      <th>ICD9_V443_count</th>\n",
       "      <th>ICD9_V444_count</th>\n",
       "      <th>ICD9_V446_count</th>\n",
       "      <th>ICD9_V4501_count</th>\n",
       "      <th>ICD9_V4502_count</th>\n",
       "      <th>ICD9_V451_count</th>\n",
       "      <th>ICD9_V4511_count</th>\n",
       "      <th>ICD9_V452_count</th>\n",
       "      <th>ICD9_V453_count</th>\n",
       "      <th>ICD9_V454_count</th>\n",
       "      <th>ICD9_V4561_count</th>\n",
       "      <th>ICD9_V4571_count</th>\n",
       "      <th>ICD9_V4572_count</th>\n",
       "      <th>ICD9_V4573_count</th>\n",
       "      <th>ICD9_V4577_count</th>\n",
       "      <th>ICD9_V4579_count</th>\n",
       "      <th>ICD9_V4581_count</th>\n",
       "      <th>ICD9_V4582_count</th>\n",
       "      <th>ICD9_V4585_count</th>\n",
       "      <th>ICD9_V4586_count</th>\n",
       "      <th>ICD9_V4588_count</th>\n",
       "      <th>ICD9_V4589_count</th>\n",
       "      <th>ICD9_V4611_count</th>\n",
       "      <th>ICD9_V462_count</th>\n",
       "      <th>ICD9_V463_count</th>\n",
       "      <th>ICD9_V4972_count</th>\n",
       "      <th>ICD9_V4975_count</th>\n",
       "      <th>ICD9_V4976_count</th>\n",
       "      <th>ICD9_V4983_count</th>\n",
       "      <th>ICD9_V4986_count</th>\n",
       "      <th>ICD9_V4987_count</th>\n",
       "      <th>ICD9_V502_count</th>\n",
       "      <th>ICD9_V5331_count</th>\n",
       "      <th>ICD9_V5332_count</th>\n",
       "      <th>ICD9_V550_count</th>\n",
       "      <th>ICD9_V551_count</th>\n",
       "      <th>ICD9_V552_count</th>\n",
       "      <th>ICD9_V553_count</th>\n",
       "      <th>ICD9_V5861_count</th>\n",
       "      <th>ICD9_V5865_count</th>\n",
       "      <th>ICD9_V5866_count</th>\n",
       "      <th>ICD9_V5867_count</th>\n",
       "      <th>ICD9_V5869_count</th>\n",
       "      <th>ICD9_V5883_count</th>\n",
       "      <th>ICD9_V600_count</th>\n",
       "      <th>ICD9_V6284_count</th>\n",
       "      <th>ICD9_V641_count</th>\n",
       "      <th>ICD9_V642_count</th>\n",
       "      <th>ICD9_V643_count</th>\n",
       "      <th>ICD9_V6441_count</th>\n",
       "      <th>ICD9_V667_count</th>\n",
       "      <th>ICD9_V707_count</th>\n",
       "      <th>ICD9_V721_count</th>\n",
       "      <th>ICD9_V850_count</th>\n",
       "      <th>ICD9_V851_count</th>\n",
       "      <th>ICD9_V854_count</th>\n",
       "      <th>ICD9_V8541_count</th>\n",
       "      <th>ICD9_V8542_count</th>\n",
       "      <th>ICD9_V8543_count</th>\n",
       "      <th>ICD9_V8741_count</th>\n",
       "      <th>ICD9_V8801_count</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 1489 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            ICD9_00845_count  ICD9_0088_count  ICD9_0380_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                        0.0              0.0              0.0   \n",
       "3                        0.0              0.0              0.0   \n",
       "4                        0.0              0.0              0.0   \n",
       "6                        0.0              0.0              0.0   \n",
       "7                        0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_03811_count  ICD9_03812_count  ICD9_03819_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_0382_count  ICD9_0383_count  ICD9_03840_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0              0.0               0.0   \n",
       "3                       0.0              0.0               0.0   \n",
       "4                       0.0              0.0               0.0   \n",
       "6                       0.0              0.0               0.0   \n",
       "7                       0.0              0.0               0.0   \n",
       "\n",
       "            ICD9_03842_count  ICD9_03843_count  ICD9_03849_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_0388_count  ICD9_0389_count  ICD9_04101_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0              0.0               0.0   \n",
       "3                       0.0              1.0               0.0   \n",
       "4                       0.0              0.0               0.0   \n",
       "6                       0.0              0.0               0.0   \n",
       "7                       0.0              0.0               0.0   \n",
       "\n",
       "            ICD9_04102_count  ICD9_04104_count  ICD9_04109_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_04110_count  ICD9_04111_count  ICD9_04112_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               1.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_04119_count  ICD9_0413_count  ICD9_0414_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                        0.0              0.0              0.0   \n",
       "3                        0.0              0.0              0.0   \n",
       "4                        0.0              0.0              0.0   \n",
       "6                        0.0              0.0              0.0   \n",
       "7                        0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_04149_count  ICD9_0415_count  ICD9_0416_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                        0.0              0.0              0.0   \n",
       "3                        0.0              0.0              0.0   \n",
       "4                        0.0              0.0              0.0   \n",
       "6                        0.0              0.0              0.0   \n",
       "7                        0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_0417_count  ICD9_04184_count  ICD9_04185_count  \\\n",
       "SUBJECT_ID                                                        \n",
       "2                       0.0               0.0               0.0   \n",
       "3                       0.0               0.0               0.0   \n",
       "4                       0.0               0.0               0.0   \n",
       "6                       0.0               0.0               0.0   \n",
       "7                       0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_04186_count  ICD9_04189_count  ICD9_042_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                        0.0               0.0             0.0   \n",
       "3                        0.0               0.0             0.0   \n",
       "4                        0.0               0.0             1.0   \n",
       "6                        0.0               0.0             0.0   \n",
       "7                        0.0               0.0             0.0   \n",
       "\n",
       "            ICD9_0539_count  ICD9_0549_count  ICD9_07030_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0              0.0               0.0   \n",
       "3                       0.0              0.0               0.0   \n",
       "4                       0.0              0.0               0.0   \n",
       "6                       0.0              0.0               0.0   \n",
       "7                       0.0              0.0               0.0   \n",
       "\n",
       "            ICD9_07032_count  ICD9_07044_count  ICD9_07051_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_07054_count  ICD9_07070_count  ICD9_07071_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_0785_count  ICD9_07999_count  ICD9_1101_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0               0.0              0.0   \n",
       "3                       0.0               0.0              0.0   \n",
       "4                       0.0               0.0              0.0   \n",
       "6                       0.0               0.0              0.0   \n",
       "7                       0.0               0.0              0.0   \n",
       "\n",
       "            ICD9_1103_count  ICD9_1120_count  ICD9_1121_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_1122_count  ICD9_1123_count  ICD9_1125_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_11284_count  ICD9_11289_count  ICD9_1173_count  \\\n",
       "SUBJECT_ID                                                        \n",
       "2                        0.0               0.0              0.0   \n",
       "3                        0.0               0.0              0.0   \n",
       "4                        0.0               0.0              0.0   \n",
       "6                        0.0               0.0              0.0   \n",
       "7                        0.0               0.0              0.0   \n",
       "\n",
       "            ICD9_1179_count  ICD9_135_count  ICD9_1363_count  ICD9_138_count  \\\n",
       "SUBJECT_ID                                                                     \n",
       "2                       0.0             0.0              0.0             0.0   \n",
       "3                       0.0             0.0              0.0             0.0   \n",
       "4                       0.0             0.0              1.0             0.0   \n",
       "6                       0.0             0.0              0.0             0.0   \n",
       "7                       0.0             0.0              0.0             0.0   \n",
       "\n",
       "            ICD9_1505_count  ICD9_1508_count  ICD9_1510_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_1541_count  ICD9_1550_count  ICD9_1551_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_1552_count  ICD9_1570_count  ICD9_1578_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_1622_count  ICD9_1623_count  ICD9_1625_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_1628_count  ICD9_1629_count  ICD9_1748_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_1749_count  ICD9_1830_count  ...  ICD9_V3101_count  \\\n",
       "SUBJECT_ID                                    ...                     \n",
       "2                       0.0              0.0  ...               0.0   \n",
       "3                       0.0              0.0  ...               0.0   \n",
       "4                       0.0              0.0  ...               0.0   \n",
       "6                       0.0              0.0  ...               0.0   \n",
       "7                       0.0              0.0  ...               0.0   \n",
       "\n",
       "            ICD9_V3401_count  ICD9_V420_count  ICD9_V422_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                        0.0              0.0              0.0   \n",
       "3                        0.0              0.0              0.0   \n",
       "4                        0.0              0.0              0.0   \n",
       "6                        0.0              0.0              0.0   \n",
       "7                        0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_V427_count  ICD9_V4281_count  ICD9_V4282_count  \\\n",
       "SUBJECT_ID                                                        \n",
       "2                       0.0               0.0               0.0   \n",
       "3                       0.0               0.0               0.0   \n",
       "4                       0.0               0.0               0.0   \n",
       "6                       0.0               0.0               0.0   \n",
       "7                       0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V4283_count  ICD9_V433_count  ICD9_V4364_count  \\\n",
       "SUBJECT_ID                                                        \n",
       "2                        0.0              0.0               0.0   \n",
       "3                        0.0              0.0               0.0   \n",
       "4                        0.0              0.0               0.0   \n",
       "6                        0.0              0.0               0.0   \n",
       "7                        0.0              0.0               0.0   \n",
       "\n",
       "            ICD9_V4365_count  ICD9_V440_count  ICD9_V441_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                        0.0              0.0              0.0   \n",
       "3                        0.0              0.0              0.0   \n",
       "4                        0.0              0.0              0.0   \n",
       "6                        0.0              0.0              0.0   \n",
       "7                        0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_V442_count  ICD9_V443_count  ICD9_V444_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_V446_count  ICD9_V4501_count  ICD9_V4502_count  \\\n",
       "SUBJECT_ID                                                        \n",
       "2                       0.0               0.0               0.0   \n",
       "3                       0.0               0.0               0.0   \n",
       "4                       0.0               0.0               0.0   \n",
       "6                       0.0               0.0               0.0   \n",
       "7                       0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V451_count  ICD9_V4511_count  ICD9_V452_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0               0.0              0.0   \n",
       "3                       0.0               0.0              0.0   \n",
       "4                       0.0               0.0              0.0   \n",
       "6                       0.0               0.0              0.0   \n",
       "7                       0.0               0.0              0.0   \n",
       "\n",
       "            ICD9_V453_count  ICD9_V454_count  ICD9_V4561_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0              0.0               0.0   \n",
       "3                       0.0              0.0               0.0   \n",
       "4                       0.0              0.0               0.0   \n",
       "6                       0.0              0.0               0.0   \n",
       "7                       0.0              0.0               0.0   \n",
       "\n",
       "            ICD9_V4571_count  ICD9_V4572_count  ICD9_V4573_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V4577_count  ICD9_V4579_count  ICD9_V4581_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V4582_count  ICD9_V4585_count  ICD9_V4586_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V4588_count  ICD9_V4589_count  ICD9_V4611_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V462_count  ICD9_V463_count  ICD9_V4972_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0              0.0               0.0   \n",
       "3                       0.0              0.0               0.0   \n",
       "4                       0.0              0.0               0.0   \n",
       "6                       0.0              0.0               0.0   \n",
       "7                       0.0              0.0               0.0   \n",
       "\n",
       "            ICD9_V4975_count  ICD9_V4976_count  ICD9_V4983_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V4986_count  ICD9_V4987_count  ICD9_V502_count  \\\n",
       "SUBJECT_ID                                                        \n",
       "2                        0.0               0.0              0.0   \n",
       "3                        0.0               0.0              0.0   \n",
       "4                        0.0               0.0              0.0   \n",
       "6                        0.0               0.0              0.0   \n",
       "7                        0.0               0.0              0.0   \n",
       "\n",
       "            ICD9_V5331_count  ICD9_V5332_count  ICD9_V550_count  \\\n",
       "SUBJECT_ID                                                        \n",
       "2                        0.0               0.0              0.0   \n",
       "3                        0.0               0.0              0.0   \n",
       "4                        0.0               0.0              0.0   \n",
       "6                        0.0               0.0              0.0   \n",
       "7                        0.0               0.0              0.0   \n",
       "\n",
       "            ICD9_V551_count  ICD9_V552_count  ICD9_V553_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_V5861_count  ICD9_V5865_count  ICD9_V5866_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V5867_count  ICD9_V5869_count  ICD9_V5883_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V600_count  ICD9_V6284_count  ICD9_V641_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0               0.0              0.0   \n",
       "3                       0.0               0.0              0.0   \n",
       "4                       0.0               0.0              0.0   \n",
       "6                       0.0               0.0              0.0   \n",
       "7                       0.0               0.0              0.0   \n",
       "\n",
       "            ICD9_V642_count  ICD9_V643_count  ICD9_V6441_count  \\\n",
       "SUBJECT_ID                                                       \n",
       "2                       0.0              0.0               0.0   \n",
       "3                       0.0              0.0               0.0   \n",
       "4                       0.0              0.0               0.0   \n",
       "6                       0.0              0.0               0.0   \n",
       "7                       0.0              0.0               0.0   \n",
       "\n",
       "            ICD9_V667_count  ICD9_V707_count  ICD9_V721_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_V850_count  ICD9_V851_count  ICD9_V854_count  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "3                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "7                       0.0              0.0              0.0   \n",
       "\n",
       "            ICD9_V8541_count  ICD9_V8542_count  ICD9_V8543_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "3                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "7                        0.0               0.0               0.0   \n",
       "\n",
       "            ICD9_V8741_count  ICD9_V8801_count  \n",
       "SUBJECT_ID                                      \n",
       "2                        0.0               0.0  \n",
       "3                        0.0               0.0  \n",
       "4                        0.0               0.0  \n",
       "6                        0.0               0.0  \n",
       "7                        0.0               0.0  \n",
       "\n",
       "[5 rows x 1489 columns]"
      ]
     },
     "execution_count": 71,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "diag_final.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total feats: 719\n",
      "Feats after leaving only train: 710\n",
      "Feats after removing rare: 476\n"
     ]
    }
   ],
   "source": [
    "labs_built = build_feats(lab, agg=['mean', 'max', 'min'], train_ids=TRAIN_IDS, low_thresh=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "labs_final = pivot_aggregation(labs_built, fill_value=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>LAB_50800_mean</th>\n",
       "      <th>LAB_50801_mean</th>\n",
       "      <th>LAB_50802_mean</th>\n",
       "      <th>LAB_50803_mean</th>\n",
       "      <th>LAB_50804_mean</th>\n",
       "      <th>LAB_50805_mean</th>\n",
       "      <th>LAB_50806_mean</th>\n",
       "      <th>LAB_50808_mean</th>\n",
       "      <th>LAB_50809_mean</th>\n",
       "      <th>LAB_50810_mean</th>\n",
       "      <th>LAB_50811_mean</th>\n",
       "      <th>LAB_50812_mean</th>\n",
       "      <th>LAB_50813_mean</th>\n",
       "      <th>LAB_50814_mean</th>\n",
       "      <th>LAB_50815_mean</th>\n",
       "      <th>LAB_50816_mean</th>\n",
       "      <th>LAB_50817_mean</th>\n",
       "      <th>LAB_50818_mean</th>\n",
       "      <th>LAB_50819_mean</th>\n",
       "      <th>LAB_50820_mean</th>\n",
       "      <th>LAB_50821_mean</th>\n",
       "      <th>LAB_50822_mean</th>\n",
       "      <th>LAB_50823_mean</th>\n",
       "      <th>LAB_50824_mean</th>\n",
       "      <th>LAB_50825_mean</th>\n",
       "      <th>LAB_50826_mean</th>\n",
       "      <th>LAB_50827_mean</th>\n",
       "      <th>LAB_50828_mean</th>\n",
       "      <th>LAB_50829_mean</th>\n",
       "      <th>LAB_50831_mean</th>\n",
       "      <th>LAB_50835_mean</th>\n",
       "      <th>LAB_50836_mean</th>\n",
       "      <th>LAB_50838_mean</th>\n",
       "      <th>LAB_50841_mean</th>\n",
       "      <th>LAB_50842_mean</th>\n",
       "      <th>LAB_50843_mean</th>\n",
       "      <th>LAB_50849_mean</th>\n",
       "      <th>LAB_50850_mean</th>\n",
       "      <th>LAB_50852_mean</th>\n",
       "      <th>LAB_50853_mean</th>\n",
       "      <th>LAB_50854_mean</th>\n",
       "      <th>LAB_50855_mean</th>\n",
       "      <th>LAB_50856_mean</th>\n",
       "      <th>LAB_50857_mean</th>\n",
       "      <th>LAB_50861_mean</th>\n",
       "      <th>LAB_50862_mean</th>\n",
       "      <th>LAB_50863_mean</th>\n",
       "      <th>LAB_50864_mean</th>\n",
       "      <th>LAB_50866_mean</th>\n",
       "      <th>LAB_50867_mean</th>\n",
       "      <th>LAB_50868_mean</th>\n",
       "      <th>LAB_50871_mean</th>\n",
       "      <th>LAB_50872_mean</th>\n",
       "      <th>LAB_50873_mean</th>\n",
       "      <th>LAB_50874_mean</th>\n",
       "      <th>LAB_50876_mean</th>\n",
       "      <th>LAB_50877_mean</th>\n",
       "      <th>LAB_50878_mean</th>\n",
       "      <th>LAB_50879_mean</th>\n",
       "      <th>LAB_50880_mean</th>\n",
       "      <th>LAB_50881_mean</th>\n",
       "      <th>LAB_50882_mean</th>\n",
       "      <th>LAB_50883_mean</th>\n",
       "      <th>LAB_50884_mean</th>\n",
       "      <th>LAB_50885_mean</th>\n",
       "      <th>LAB_50886_mean</th>\n",
       "      <th>LAB_50887_mean</th>\n",
       "      <th>LAB_50889_mean</th>\n",
       "      <th>LAB_50890_mean</th>\n",
       "      <th>LAB_50891_mean</th>\n",
       "      <th>LAB_50892_mean</th>\n",
       "      <th>LAB_50893_mean</th>\n",
       "      <th>LAB_50895_mean</th>\n",
       "      <th>LAB_50896_mean</th>\n",
       "      <th>LAB_50898_mean</th>\n",
       "      <th>...</th>\n",
       "      <th>LAB_51404_min</th>\n",
       "      <th>LAB_51406_min</th>\n",
       "      <th>LAB_51411_min</th>\n",
       "      <th>LAB_51412_min</th>\n",
       "      <th>LAB_51416_min</th>\n",
       "      <th>LAB_51419_min</th>\n",
       "      <th>LAB_51420_min</th>\n",
       "      <th>LAB_51422_min</th>\n",
       "      <th>LAB_51423_min</th>\n",
       "      <th>LAB_51424_min</th>\n",
       "      <th>LAB_51425_min</th>\n",
       "      <th>LAB_51426_min</th>\n",
       "      <th>LAB_51427_min</th>\n",
       "      <th>LAB_51428_min</th>\n",
       "      <th>LAB_51429_min</th>\n",
       "      <th>LAB_51431_min</th>\n",
       "      <th>LAB_51434_min</th>\n",
       "      <th>LAB_51436_min</th>\n",
       "      <th>LAB_51438_min</th>\n",
       "      <th>LAB_51439_min</th>\n",
       "      <th>LAB_51440_min</th>\n",
       "      <th>LAB_51441_min</th>\n",
       "      <th>LAB_51442_min</th>\n",
       "      <th>LAB_51444_min</th>\n",
       "      <th>LAB_51445_min</th>\n",
       "      <th>LAB_51446_min</th>\n",
       "      <th>LAB_51447_min</th>\n",
       "      <th>LAB_51448_min</th>\n",
       "      <th>LAB_51450_min</th>\n",
       "      <th>LAB_51453_min</th>\n",
       "      <th>LAB_51454_min</th>\n",
       "      <th>LAB_51455_min</th>\n",
       "      <th>LAB_51457_min</th>\n",
       "      <th>LAB_51458_min</th>\n",
       "      <th>LAB_51460_min</th>\n",
       "      <th>LAB_51462_min</th>\n",
       "      <th>LAB_51463_min</th>\n",
       "      <th>LAB_51464_min</th>\n",
       "      <th>LAB_51465_min</th>\n",
       "      <th>LAB_51466_min</th>\n",
       "      <th>LAB_51469_min</th>\n",
       "      <th>LAB_51471_min</th>\n",
       "      <th>LAB_51474_min</th>\n",
       "      <th>LAB_51476_min</th>\n",
       "      <th>LAB_51478_min</th>\n",
       "      <th>LAB_51479_min</th>\n",
       "      <th>LAB_51481_min</th>\n",
       "      <th>LAB_51482_min</th>\n",
       "      <th>LAB_51484_min</th>\n",
       "      <th>LAB_51486_min</th>\n",
       "      <th>LAB_51487_min</th>\n",
       "      <th>LAB_51489_min</th>\n",
       "      <th>LAB_51491_min</th>\n",
       "      <th>LAB_51492_min</th>\n",
       "      <th>LAB_51493_min</th>\n",
       "      <th>LAB_51497_min</th>\n",
       "      <th>LAB_51498_min</th>\n",
       "      <th>LAB_51499_min</th>\n",
       "      <th>LAB_51501_min</th>\n",
       "      <th>LAB_51503_min</th>\n",
       "      <th>LAB_51505_min</th>\n",
       "      <th>LAB_51506_min</th>\n",
       "      <th>LAB_51507_min</th>\n",
       "      <th>LAB_51508_min</th>\n",
       "      <th>LAB_51510_min</th>\n",
       "      <th>LAB_51512_min</th>\n",
       "      <th>LAB_51513_min</th>\n",
       "      <th>LAB_51514_min</th>\n",
       "      <th>LAB_51515_min</th>\n",
       "      <th>LAB_51516_min</th>\n",
       "      <th>LAB_51517_min</th>\n",
       "      <th>LAB_51518_min</th>\n",
       "      <th>LAB_51519_min</th>\n",
       "      <th>LAB_51523_min</th>\n",
       "      <th>LAB_51529_min</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.300000</td>\n",
       "      <td>9.000000</td>\n",
       "      <td>9.30</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.0</td>\n",
       "      <td>530.0</td>\n",
       "      <td>-4.266667</td>\n",
       "      <td>20.0</td>\n",
       "      <td>19.844444</td>\n",
       "      <td>0.0</td>\n",
       "      <td>108.4</td>\n",
       "      <td>1.090417</td>\n",
       "      <td>174.333333</td>\n",
       "      <td>30.173913</td>\n",
       "      <td>10.065217</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.841176</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>41.785714</td>\n",
       "      <td>82.475</td>\n",
       "      <td>31.6</td>\n",
       "      <td>6.32</td>\n",
       "      <td>7.388958</td>\n",
       "      <td>114.955556</td>\n",
       "      <td>4.246154</td>\n",
       "      <td>85.0</td>\n",
       "      <td>140.222222</td>\n",
       "      <td>37.089286</td>\n",
       "      <td>607.666667</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>18.000000</td>\n",
       "      <td>2.300000</td>\n",
       "      <td>81.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>35.000000</td>\n",
       "      <td>14.705882</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>56.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>22.085714</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.80</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.925926</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>100.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>500.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.014</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>10.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-3.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>19.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>32.000000</td>\n",
       "      <td>10.500000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.580000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>40.000000</td>\n",
       "      <td>94.000</td>\n",
       "      <td>27.5</td>\n",
       "      <td>0.00</td>\n",
       "      <td>7.396667</td>\n",
       "      <td>58.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>38.900000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.4</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>40.658537</td>\n",
       "      <td>3.743750</td>\n",
       "      <td>1195.421053</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>81.714286</td>\n",
       "      <td>16.217391</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>63.552632</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>23.608696</td>\n",
       "      <td>1.866667</td>\n",
       "      <td>0.966667</td>\n",
       "      <td>2.55</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>9.440741</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.005</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>-4.400000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>21.200000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>100.8</td>\n",
       "      <td>1.150000</td>\n",
       "      <td>144.400000</td>\n",
       "      <td>27.400000</td>\n",
       "      <td>9.080000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.640000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>37.2</td>\n",
       "      <td>0.00</td>\n",
       "      <td>7.342000</td>\n",
       "      <td>127.400000</td>\n",
       "      <td>4.240000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>135.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>14.533333</td>\n",
       "      <td>3.857143</td>\n",
       "      <td>82.750000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>16.406250</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>27.562500</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>20.625000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.25</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>8.748148</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>30.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.010</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.000</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 1428 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            LAB_50800_mean  LAB_50801_mean  LAB_50802_mean  LAB_50803_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0             0.0        0.000000             0.0   \n",
       "3                      0.0           530.0       -4.266667            20.0   \n",
       "4                      0.0             0.0       -3.000000             0.0   \n",
       "6                      0.0             0.0       -4.400000             0.0   \n",
       "7                      0.0             0.0        0.000000             0.0   \n",
       "\n",
       "            LAB_50804_mean  LAB_50805_mean  LAB_50806_mean  LAB_50808_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                 0.000000             0.0             0.0        0.000000   \n",
       "3                19.844444             0.0           108.4        1.090417   \n",
       "4                19.500000             0.0             0.0        0.000000   \n",
       "6                21.200000             0.0           100.8        1.150000   \n",
       "7                 0.000000             0.0             0.0        0.000000   \n",
       "\n",
       "            LAB_50809_mean  LAB_50810_mean  LAB_50811_mean  LAB_50812_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                 0.000000        0.000000        0.000000             0.0   \n",
       "3               174.333333       30.173913       10.065217             0.0   \n",
       "4               100.000000       32.000000       10.500000             0.0   \n",
       "6               144.400000       27.400000        9.080000             0.0   \n",
       "7                 0.000000        0.000000        0.000000             0.0   \n",
       "\n",
       "            LAB_50813_mean  LAB_50814_mean  LAB_50815_mean  LAB_50816_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                 0.000000             0.0             0.0        0.000000   \n",
       "3                 2.841176             0.0             0.0       41.785714   \n",
       "4                 2.580000             0.0             6.0       40.000000   \n",
       "6                 2.640000             0.0             0.0        0.000000   \n",
       "7                 0.000000             0.0             0.0        0.000000   \n",
       "\n",
       "            LAB_50817_mean  LAB_50818_mean  LAB_50819_mean  LAB_50820_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                    0.000             0.0            0.00        0.000000   \n",
       "3                   82.475            31.6            6.32        7.388958   \n",
       "4                   94.000            27.5            0.00        7.396667   \n",
       "6                    0.000            37.2            0.00        7.342000   \n",
       "7                    0.000             0.0            0.00        0.000000   \n",
       "\n",
       "            LAB_50821_mean  LAB_50822_mean  LAB_50823_mean  LAB_50824_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                 0.000000        0.000000             0.0        0.000000   \n",
       "3               114.955556        4.246154            85.0      140.222222   \n",
       "4                58.000000        0.000000             0.0        0.000000   \n",
       "6               127.400000        4.240000             0.0      135.000000   \n",
       "7                 0.000000        0.000000             0.0        0.000000   \n",
       "\n",
       "            LAB_50825_mean  LAB_50826_mean  LAB_50827_mean  LAB_50828_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                 0.000000        0.000000             0.0             0.0   \n",
       "3                37.089286      607.666667             0.0             0.0   \n",
       "4                38.900000        0.000000             0.0             0.0   \n",
       "6                 0.000000        0.000000             0.0             0.0   \n",
       "7                 0.000000        0.000000             0.0             0.0   \n",
       "\n",
       "            LAB_50829_mean  LAB_50831_mean  LAB_50835_mean  LAB_50836_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0             0.0             0.0             0.0   \n",
       "3                      0.0             0.0             0.0             0.0   \n",
       "4                      0.0             0.0             0.0             0.0   \n",
       "6                      0.0             0.0             0.0             0.0   \n",
       "7                      0.0             0.0             0.0             0.0   \n",
       "\n",
       "            LAB_50838_mean  LAB_50841_mean  LAB_50842_mean  LAB_50843_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0             0.0             0.0             0.0   \n",
       "3                      0.0             0.0             0.0             0.0   \n",
       "4                      0.0             0.0             0.0             0.0   \n",
       "6                      0.0             0.0             0.0             0.0   \n",
       "7                      0.0             0.0             0.0             0.0   \n",
       "\n",
       "            LAB_50849_mean  LAB_50850_mean  LAB_50852_mean  LAB_50853_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0             0.0             0.0             0.0   \n",
       "3                      0.0             0.0             0.0             0.0   \n",
       "4                      0.0             0.0             7.4             0.0   \n",
       "6                      0.0             0.0             0.0             0.0   \n",
       "7                      0.0             0.0             0.0             0.0   \n",
       "\n",
       "            LAB_50854_mean  LAB_50855_mean  LAB_50856_mean  LAB_50857_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0             0.0             0.0             0.0   \n",
       "3                      0.0             0.0             0.0             0.0   \n",
       "4                      0.0             0.0             0.0             0.0   \n",
       "6                      0.0             0.0             0.0             0.0   \n",
       "7                      0.0             0.0             0.0             0.0   \n",
       "\n",
       "            LAB_50861_mean  LAB_50862_mean  LAB_50863_mean  LAB_50864_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                 0.000000        0.000000        0.000000             0.0   \n",
       "3                18.000000        2.300000       81.000000             0.0   \n",
       "4                40.658537        3.743750     1195.421053             0.0   \n",
       "6                14.533333        3.857143       82.750000             0.0   \n",
       "7                 0.000000        0.000000        0.000000             0.0   \n",
       "\n",
       "            LAB_50866_mean  LAB_50867_mean  LAB_50868_mean  LAB_50871_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0        0.000000        0.000000             0.0   \n",
       "3                      0.0       35.000000       14.705882             0.0   \n",
       "4                      0.0       81.714286       16.217391             0.0   \n",
       "6                      0.0        0.000000       16.406250             0.0   \n",
       "7                      0.0        0.000000        0.000000             0.0   \n",
       "\n",
       "            LAB_50872_mean  LAB_50873_mean  LAB_50874_mean  LAB_50876_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0             0.0             0.0             0.0   \n",
       "3                      0.0             0.0             0.0             0.0   \n",
       "4                      0.0             0.0             0.0             0.0   \n",
       "6                      0.0             0.0             0.0             0.0   \n",
       "7                      0.0             0.0             0.0             0.0   \n",
       "\n",
       "            LAB_50877_mean  LAB_50878_mean  LAB_50879_mean  LAB_50880_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0        0.000000             0.0             0.0   \n",
       "3                      0.0       56.000000             0.0             0.0   \n",
       "4                      0.0       63.552632             0.0             0.0   \n",
       "6                      0.0       27.562500             0.0             0.0   \n",
       "7                      0.0        0.000000             0.0             0.0   \n",
       "\n",
       "            LAB_50881_mean  LAB_50882_mean  LAB_50883_mean  LAB_50884_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0        0.000000        0.300000        9.000000   \n",
       "3                      0.0       22.085714        0.000000        0.000000   \n",
       "4                      0.0       23.608696        1.866667        0.966667   \n",
       "6                      0.0       20.625000        0.000000        0.000000   \n",
       "7                      0.0        0.000000        0.000000        0.000000   \n",
       "\n",
       "            LAB_50885_mean  LAB_50886_mean  LAB_50887_mean  LAB_50889_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                     9.30             0.0             0.0             0.0   \n",
       "3                     0.80             0.0             0.0             0.0   \n",
       "4                     2.55             0.0             0.0             0.0   \n",
       "6                     0.25             0.0             0.0             0.2   \n",
       "7                     0.00             0.0             0.0             0.0   \n",
       "\n",
       "            LAB_50890_mean  LAB_50891_mean  LAB_50892_mean  LAB_50893_mean  \\\n",
       "SUBJECT_ID                                                                   \n",
       "2                      0.0             0.0             0.0        0.000000   \n",
       "3                      0.0             0.0             0.0        7.925926   \n",
       "4                      0.0             0.0             0.0        9.440741   \n",
       "6                      0.0             0.0             0.0        8.748148   \n",
       "7                      0.0             0.0             0.0        0.000000   \n",
       "\n",
       "            LAB_50895_mean  LAB_50896_mean  LAB_50898_mean  ...  \\\n",
       "SUBJECT_ID                                                  ...   \n",
       "2                      0.0             0.0             0.0  ...   \n",
       "3                      0.0             0.0             0.0  ...   \n",
       "4                      0.0             0.0             0.0  ...   \n",
       "6                      0.0             0.0             0.0  ...   \n",
       "7                      0.0             0.0             0.0  ...   \n",
       "\n",
       "            LAB_51404_min  LAB_51406_min  LAB_51411_min  LAB_51412_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51416_min  LAB_51419_min  LAB_51420_min  LAB_51422_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51423_min  LAB_51424_min  LAB_51425_min  LAB_51426_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51427_min  LAB_51428_min  LAB_51429_min  LAB_51431_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51434_min  LAB_51436_min  LAB_51438_min  LAB_51439_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51440_min  LAB_51441_min  LAB_51442_min  LAB_51444_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51445_min  LAB_51446_min  LAB_51447_min  LAB_51448_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51450_min  LAB_51453_min  LAB_51454_min  LAB_51455_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51457_min  LAB_51458_min  LAB_51460_min  LAB_51462_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51463_min  LAB_51464_min  LAB_51465_min  LAB_51466_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51469_min  LAB_51471_min  LAB_51474_min  LAB_51476_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51478_min  LAB_51479_min  LAB_51481_min  LAB_51482_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                   100.0            0.0            0.0            4.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51484_min  LAB_51486_min  LAB_51487_min  LAB_51489_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51491_min  LAB_51492_min  LAB_51493_min  LAB_51497_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     5.0          500.0            0.0            0.0   \n",
       "4                     5.0           30.0            0.0            0.0   \n",
       "6                     5.0           30.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51498_min  LAB_51499_min  LAB_51501_min  LAB_51503_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                   0.000            0.0            0.0            0.0   \n",
       "3                   1.014            0.0            0.0            0.0   \n",
       "4                   1.005            0.0            0.0            0.0   \n",
       "6                   1.010            0.0            0.0            0.0   \n",
       "7                   0.000            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51505_min  LAB_51506_min  LAB_51507_min  LAB_51508_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51510_min  LAB_51512_min  LAB_51513_min  LAB_51514_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0            0.0            0.0            0.2   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            0.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51515_min  LAB_51516_min  LAB_51517_min  LAB_51518_min  \\\n",
       "SUBJECT_ID                                                               \n",
       "2                     0.0            0.0            0.0            0.0   \n",
       "3                     0.0           10.0            0.0            0.0   \n",
       "4                     0.0            0.0            0.0            0.0   \n",
       "6                     0.0            2.0            0.0            0.0   \n",
       "7                     0.0            0.0            0.0            0.0   \n",
       "\n",
       "            LAB_51519_min  LAB_51523_min  LAB_51529_min  \n",
       "SUBJECT_ID                                               \n",
       "2                     0.0            0.0            0.0  \n",
       "3                     0.0            0.0            0.0  \n",
       "4                     0.0            0.0            0.0  \n",
       "6                     0.0            0.0            0.0  \n",
       "7                     0.0            0.0            0.0  \n",
       "\n",
       "[5 rows x 1428 columns]"
      ]
     },
     "execution_count": 74,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "labs_final.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total feats: 4998\n",
      "Feats after leaving only train: 4806\n",
      "Feats after removing rare: 1601\n"
     ]
    }
   ],
   "source": [
    "meds_built = build_feats(med, agg=['mean', 'count'], train_ids=TRAIN_IDS, low_thresh=50)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "meds_final = pivot_aggregation(meds_built, fill_value=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>MED_000015_mean</th>\n",
       "      <th>MED_000018_mean</th>\n",
       "      <th>MED_000019_mean</th>\n",
       "      <th>MED_000175_mean</th>\n",
       "      <th>MED_000246_mean</th>\n",
       "      <th>MED_000253_mean</th>\n",
       "      <th>MED_000266_mean</th>\n",
       "      <th>MED_000283_mean</th>\n",
       "      <th>MED_000284_mean</th>\n",
       "      <th>MED_000286_mean</th>\n",
       "      <th>MED_000287_mean</th>\n",
       "      <th>MED_000343_mean</th>\n",
       "      <th>MED_000344_mean</th>\n",
       "      <th>MED_000345_mean</th>\n",
       "      <th>MED_000346_mean</th>\n",
       "      <th>MED_000347_mean</th>\n",
       "      <th>MED_000379_mean</th>\n",
       "      <th>MED_000380_mean</th>\n",
       "      <th>MED_000383_mean</th>\n",
       "      <th>MED_000384_mean</th>\n",
       "      <th>MED_000387_mean</th>\n",
       "      <th>MED_000390_mean</th>\n",
       "      <th>MED_000391_mean</th>\n",
       "      <th>MED_000393_mean</th>\n",
       "      <th>MED_000463_mean</th>\n",
       "      <th>MED_000464_mean</th>\n",
       "      <th>MED_000474_mean</th>\n",
       "      <th>MED_000475_mean</th>\n",
       "      <th>MED_000507_mean</th>\n",
       "      <th>MED_000508_mean</th>\n",
       "      <th>MED_000564_mean</th>\n",
       "      <th>MED_000565_mean</th>\n",
       "      <th>MED_000566_mean</th>\n",
       "      <th>MED_000567_mean</th>\n",
       "      <th>MED_000568_mean</th>\n",
       "      <th>MED_000574_mean</th>\n",
       "      <th>MED_000575_mean</th>\n",
       "      <th>MED_000576_mean</th>\n",
       "      <th>MED_000579_mean</th>\n",
       "      <th>MED_000591_mean</th>\n",
       "      <th>MED_000592_mean</th>\n",
       "      <th>MED_000652_mean</th>\n",
       "      <th>MED_000759_mean</th>\n",
       "      <th>MED_000762_mean</th>\n",
       "      <th>MED_001166_mean</th>\n",
       "      <th>MED_001181_mean</th>\n",
       "      <th>MED_001183_mean</th>\n",
       "      <th>MED_001185_mean</th>\n",
       "      <th>MED_001187_mean</th>\n",
       "      <th>MED_001195_mean</th>\n",
       "      <th>MED_001198_mean</th>\n",
       "      <th>MED_001205_mean</th>\n",
       "      <th>MED_001209_mean</th>\n",
       "      <th>MED_001210_mean</th>\n",
       "      <th>MED_001211_mean</th>\n",
       "      <th>MED_001220_mean</th>\n",
       "      <th>MED_001226_mean</th>\n",
       "      <th>MED_001233_mean</th>\n",
       "      <th>MED_001241_mean</th>\n",
       "      <th>MED_001248_mean</th>\n",
       "      <th>MED_001255_mean</th>\n",
       "      <th>MED_001262_mean</th>\n",
       "      <th>MED_001275_mean</th>\n",
       "      <th>MED_001285_mean</th>\n",
       "      <th>MED_001340_mean</th>\n",
       "      <th>MED_001349_mean</th>\n",
       "      <th>MED_001356_mean</th>\n",
       "      <th>MED_001405_mean</th>\n",
       "      <th>MED_001408_mean</th>\n",
       "      <th>MED_001412_mean</th>\n",
       "      <th>MED_001417_mean</th>\n",
       "      <th>MED_001627_mean</th>\n",
       "      <th>MED_001640_mean</th>\n",
       "      <th>MED_001641_mean</th>\n",
       "      <th>MED_001645_mean</th>\n",
       "      <th>...</th>\n",
       "      <th>MED_7894.0_count</th>\n",
       "      <th>MED_7936.0_count</th>\n",
       "      <th>MED_7948.0_count</th>\n",
       "      <th>MED_7993.0_count</th>\n",
       "      <th>MED_8039.0_count</th>\n",
       "      <th>MED_8084.0_count</th>\n",
       "      <th>MED_8148.0_count</th>\n",
       "      <th>MED_8163.0_count</th>\n",
       "      <th>MED_8166.0_count</th>\n",
       "      <th>MED_8171.0_count</th>\n",
       "      <th>MED_8182.0_count</th>\n",
       "      <th>MED_8205.0_count</th>\n",
       "      <th>MED_8207.0_count</th>\n",
       "      <th>MED_8208.0_count</th>\n",
       "      <th>MED_8209.0_count</th>\n",
       "      <th>MED_8210.0_count</th>\n",
       "      <th>MED_8217.0_count</th>\n",
       "      <th>MED_8218.0_count</th>\n",
       "      <th>MED_8220.0_count</th>\n",
       "      <th>MED_8223.0_count</th>\n",
       "      <th>MED_8270.0_count</th>\n",
       "      <th>MED_8334.0_count</th>\n",
       "      <th>MED_8335.0_count</th>\n",
       "      <th>MED_8336.0_count</th>\n",
       "      <th>MED_8348.0_count</th>\n",
       "      <th>MED_8349.0_count</th>\n",
       "      <th>MED_8360.0_count</th>\n",
       "      <th>MED_8398.0_count</th>\n",
       "      <th>MED_8909.0_count</th>\n",
       "      <th>MED_8920.0_count</th>\n",
       "      <th>MED_8921.0_count</th>\n",
       "      <th>MED_8932.0_count</th>\n",
       "      <th>MED_8935.0_count</th>\n",
       "      <th>MED_8937.0_count</th>\n",
       "      <th>MED_8992.0_count</th>\n",
       "      <th>MED_8995.0_count</th>\n",
       "      <th>MED_9043.0_count</th>\n",
       "      <th>MED_9061.0_count</th>\n",
       "      <th>MED_9066.0_count</th>\n",
       "      <th>MED_9156.0_count</th>\n",
       "      <th>MED_9157.0_count</th>\n",
       "      <th>MED_9172.0_count</th>\n",
       "      <th>MED_9218.0_count</th>\n",
       "      <th>MED_9221.0_count</th>\n",
       "      <th>MED_9252.0_count</th>\n",
       "      <th>MED_9263.0_count</th>\n",
       "      <th>MED_9289.0_count</th>\n",
       "      <th>MED_9291.0_count</th>\n",
       "      <th>MED_9294.0_count</th>\n",
       "      <th>MED_9298.0_count</th>\n",
       "      <th>MED_9299.0_count</th>\n",
       "      <th>MED_9310.0_count</th>\n",
       "      <th>MED_9328.0_count</th>\n",
       "      <th>MED_9329.0_count</th>\n",
       "      <th>MED_9331.0_count</th>\n",
       "      <th>MED_9339.0_count</th>\n",
       "      <th>MED_9344.0_count</th>\n",
       "      <th>MED_9361.0_count</th>\n",
       "      <th>MED_9365.0_count</th>\n",
       "      <th>MED_9393.0_count</th>\n",
       "      <th>MED_9394.0_count</th>\n",
       "      <th>MED_9395.0_count</th>\n",
       "      <th>MED_9396.0_count</th>\n",
       "      <th>MED_9477.0_count</th>\n",
       "      <th>MED_9509.0_count</th>\n",
       "      <th>MED_9510.0_count</th>\n",
       "      <th>MED_9537.0_count</th>\n",
       "      <th>MED_9553.0_count</th>\n",
       "      <th>MED_9588.0_count</th>\n",
       "      <th>MED_9592.0_count</th>\n",
       "      <th>MED_9629.0_count</th>\n",
       "      <th>MED_9630.0_count</th>\n",
       "      <th>MED_9700.0_count</th>\n",
       "      <th>MED_9839.0_count</th>\n",
       "      <th>MED_nan_count</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.5</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>250.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1000.0</td>\n",
       "      <td>531.25</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>2.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>24.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1000.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>240.00</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>...</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>13.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 3202 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            MED_000015_mean  MED_000018_mean  MED_000019_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000175_mean  MED_000246_mean  MED_000253_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000266_mean  MED_000283_mean  MED_000284_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000286_mean  MED_000287_mean  MED_000343_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000344_mean  MED_000345_mean  MED_000346_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000347_mean  MED_000379_mean  MED_000380_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000383_mean  MED_000384_mean  MED_000387_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000390_mean  MED_000391_mean  MED_000393_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000463_mean  MED_000464_mean  MED_000474_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000475_mean  MED_000507_mean  MED_000508_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000564_mean  MED_000565_mean  MED_000566_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000567_mean  MED_000568_mean  MED_000574_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000575_mean  MED_000576_mean  MED_000579_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000591_mean  MED_000592_mean  MED_000652_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              7.5   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_000759_mean  MED_000762_mean  MED_001166_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_001181_mean  MED_001183_mean  MED_001185_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_001187_mean  MED_001195_mean  MED_001198_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0           1000.0   \n",
       "\n",
       "            MED_001205_mean  MED_001209_mean  MED_001210_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0             0.00   \n",
       "4                       0.0              0.0           250.00   \n",
       "6                       0.0           1000.0           531.25   \n",
       "10                      0.0              0.0             0.00   \n",
       "11                      0.0              0.0           240.00   \n",
       "\n",
       "            MED_001211_mean  MED_001220_mean  MED_001226_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              5.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_001233_mean  MED_001241_mean  MED_001248_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0             50.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0             42.0   \n",
       "\n",
       "            MED_001255_mean  MED_001262_mean  MED_001275_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_001285_mean  MED_001340_mean  MED_001349_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       0.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_001356_mean  MED_001405_mean  MED_001408_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       0.0              0.0              0.0   \n",
       "6                       2.0              0.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_001412_mean  MED_001417_mean  MED_001627_mean  \\\n",
       "SUBJECT_ID                                                      \n",
       "2                       0.0              0.0              0.0   \n",
       "4                       1.0              0.0              0.0   \n",
       "6                       0.0              2.0              0.0   \n",
       "10                      0.0              0.0              0.0   \n",
       "11                      0.0              0.0              0.0   \n",
       "\n",
       "            MED_001640_mean  MED_001641_mean  MED_001645_mean  ...  \\\n",
       "SUBJECT_ID                                                     ...   \n",
       "2                       0.0              0.0              0.0  ...   \n",
       "4                       0.0              0.0              0.0  ...   \n",
       "6                       0.0              0.0              0.0  ...   \n",
       "10                      0.0              0.0              0.0  ...   \n",
       "11                      0.0              0.0              0.0  ...   \n",
       "\n",
       "            MED_7894.0_count  MED_7936.0_count  MED_7948.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_7993.0_count  MED_8039.0_count  MED_8084.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8148.0_count  MED_8163.0_count  MED_8166.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8171.0_count  MED_8182.0_count  MED_8205.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8207.0_count  MED_8208.0_count  MED_8209.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8210.0_count  MED_8217.0_count  MED_8218.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8220.0_count  MED_8223.0_count  MED_8270.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8334.0_count  MED_8335.0_count  MED_8336.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8348.0_count  MED_8349.0_count  MED_8360.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8398.0_count  MED_8909.0_count  MED_8920.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8921.0_count  MED_8932.0_count  MED_8935.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_8937.0_count  MED_8992.0_count  MED_8995.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9043.0_count  MED_9061.0_count  MED_9066.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9156.0_count  MED_9157.0_count  MED_9172.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9218.0_count  MED_9221.0_count  MED_9252.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9263.0_count  MED_9289.0_count  MED_9291.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9294.0_count  MED_9298.0_count  MED_9299.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9310.0_count  MED_9328.0_count  MED_9329.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9331.0_count  MED_9339.0_count  MED_9344.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9361.0_count  MED_9365.0_count  MED_9393.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9394.0_count  MED_9395.0_count  MED_9396.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9477.0_count  MED_9509.0_count  MED_9510.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9537.0_count  MED_9553.0_count  MED_9588.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9592.0_count  MED_9629.0_count  MED_9630.0_count  \\\n",
       "SUBJECT_ID                                                         \n",
       "2                        0.0               0.0               0.0   \n",
       "4                        0.0               0.0               0.0   \n",
       "6                        0.0               0.0               0.0   \n",
       "10                       0.0               0.0               0.0   \n",
       "11                       0.0               0.0               0.0   \n",
       "\n",
       "            MED_9700.0_count  MED_9839.0_count  MED_nan_count  \n",
       "SUBJECT_ID                                                     \n",
       "2                        0.0               0.0            1.0  \n",
       "4                        0.0               0.0            2.0  \n",
       "6                        0.0               0.0           24.0  \n",
       "10                       0.0               0.0            4.0  \n",
       "11                       0.0               0.0           13.0  \n",
       "\n",
       "[5 rows x 3202 columns]"
      ]
     },
     "execution_count": 77,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "meds_final.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We will use random forest to automatically incorporate feature interrelations into our model."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.feature_selection import SelectFromModel\n",
    "from sklearn.base import clone"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_df_target(*baseline_feats, feats_to_merge_left: List[pd.DataFrame] = None):\n",
    "    \"\"\"Concatenate baseline dataset and left-merge additional features.\n",
    "    \n",
    "    All datasets should have SUBJECT_ID as index.\n",
    "    \"\"\"\n",
    "    df_baseline = pd.concat(baseline_feats, axis=1).fillna(0)\n",
    "    target = pd.Series(df_baseline.index.isin(DECEASED_TO_DATE), index=df_baseline.index, name='target')\n",
    "    if feats_to_merge_left is None:\n",
    "        return df_baseline, target\n",
    "    else:\n",
    "        merge_feats = pd.concat(feats_to_merge_left, axis=1)\n",
    "        df_final = df_baseline.merge(merge_feats, left_index=True, right_index=True, how='left').fillna(0)\n",
    "        return df_final, target"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_and_predict(df: pd.DataFrame, target: pd.Series, classifier) -> np.array:\n",
    "    train_loc = df.index.isin(TRAIN_IDS)\n",
    "    try:\n",
    "        classifier.fit(df[train_loc], target[train_loc])\n",
    "        pred = classifier.predict_proba(df[~train_loc])[:, 1]\n",
    "    except AttributeError:\n",
    "        classifier.fit(df[train_loc].sparse.to_dense(), target[train_loc])\n",
    "        pred = classifier.predict_proba(df[~train_loc].sparse.to_dense())[:, 1]\n",
    "    return pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 250,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_and_predict_with_select_feats(df: pd.DataFrame, target: pd.Series, classifier, threshold=-np.inf, max_feats=500) -> np.array:\n",
    "    \"\"\"Select top features from the model and retrain the model on those.\"\"\"\n",
    "    try:\n",
    "        df = df.sparse.to_dense()\n",
    "    except AttributeError:\n",
    "        pass\n",
    "    cl_to_retrain = clone(classifier)\n",
    "    train_loc = df.index.isin(TRAIN_IDS)\n",
    "    fs = SelectFromModel(classifier, threshold=threshold, max_features=max_feats, prefit=True)\n",
    "    df_select = fs.transform(df[train_loc])\n",
    "    print(f\"Old shape: {df[train_loc].shape}, transformed shape: {df_select.shape}\")\n",
    "    cl_to_retrain.fit(df_select, target[train_loc])\n",
    "    df_test = fs.transform(df[~train_loc])\n",
    "    pred = cl_to_retrain.predict_proba(df_test)[:, 1]\n",
    "    return pred, cl_to_retrain, df.columns[fs.get_support()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 224,
   "metadata": {},
   "outputs": [],
   "source": [
    "def show_results(y_true, y_pred_proba, classifier,\n",
    "                 show_top_feats: Optional[int] = None, \n",
    "                 col_names: List[str] = None) -> None:\n",
    "    \"\"\"Print ROC-AUC score and possibly top feature importances.\"\"\"\n",
    "    print(f\"ROC-AUC: {roc_auc_score(y_true, y_pred_proba):.3f}\")\n",
    "    if show_top_feats is not None:\n",
    "        top_feats = pd.Series(classifier.feature_importances_, index=col_names)\\\n",
    "            .sort_values(ascending=False).iloc[:show_top_feats]\n",
    "        print()\n",
    "        print(top_feats)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [],
   "source": [
    "df, target = build_df_target(diag_final, meds_final, labs_final)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False    29631\n",
       "True      9280\n",
       "Name: target, dtype: int64"
      ]
     },
     "execution_count": 189,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [],
   "source": [
    "cl = RandomForestClassifier(random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 1min 21s, sys: 1.62 s, total: 1min 23s\n",
      "Wall time: 1min 24s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred = train_and_predict(df, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.884\n",
      "\n",
      "LAB_51006_mean    0.008716\n",
      "LAB_51006_max     0.008598\n",
      "LAB_51277_max     0.007897\n",
      "LAB_51277_mean    0.007630\n",
      "LAB_51249_min     0.006224\n",
      "LAB_50912_mean    0.005992\n",
      "LAB_51277_min     0.005560\n",
      "LAB_50912_max     0.005345\n",
      "LAB_50931_min     0.004884\n",
      "LAB_51006_min     0.004827\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, cl, show_top_feats=10, col_names=df.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top 500 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 219,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 6119), transformed shape: (31165, 500)\n",
      "CPU times: user 39.3 s, sys: 2.61 s, total: 42 s\n",
      "Wall time: 42.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 225,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.881\n",
      "\n",
      "LAB_51006_max      0.020538\n",
      "LAB_51006_mean     0.014755\n",
      "LAB_51277_max      0.013400\n",
      "LAB_51277_mean     0.012860\n",
      "LAB_51277_min      0.010256\n",
      "LAB_50912_max      0.009337\n",
      "ICD9_4280_count    0.009314\n",
      "LAB_51249_min      0.008912\n",
      "LAB_50931_mean     0.007554\n",
      "LAB_51006_min      0.007394\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top features based on selection strategy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 229,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 6119), transformed shape: (31165, 953)\n",
      "CPU times: user 44.2 s, sys: 2.44 s, total: 46.7 s\n",
      "Wall time: 47.3 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df, target, cl, threshold=None, max_feats=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 230,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.883\n",
      "\n",
      "LAB_51006_max     0.017249\n",
      "LAB_51006_mean    0.013202\n",
      "LAB_51277_max     0.012632\n",
      "LAB_51277_mean    0.012241\n",
      "LAB_50912_max     0.008671\n",
      "LAB_51277_min     0.007905\n",
      "LAB_50912_mean    0.007659\n",
      "LAB_51249_min     0.007013\n",
      "LAB_50931_max     0.006588\n",
      "LAB_51274_max     0.005999\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add note TF-IDF"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "last_note = clean_up_feature_sets(notes, earliest_date=earliest_date, last_date=last_date)[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols = ['SUBJECT_ID', 'DATE', 'CLEAN_TEXT']\n",
    "last_note = last_note.sort_values(by=cols, ascending=False).drop_duplicates('SUBJECT_ID')[cols]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [],
   "source": [
    "last_note.to_pickle('../data/processed/last_note.pkl')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [],
   "source": [
    "del notes"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import gc\n",
    "gc.collect()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "last_note = last_note[last_note.SUBJECT_ID.isin(SAMPLE_IDS)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>DATE</th>\n",
       "      <th>CLEAN_TEXT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>718850</th>\n",
       "      <td>99999</td>\n",
       "      <td>2118-01-01</td>\n",
       "      <td>sicu hpi 63 f pod1 s p plif with respiratory d...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1266755</th>\n",
       "      <td>99995</td>\n",
       "      <td>2147-03-16</td>\n",
       "      <td>2147 3 16 11 24 am cta abd pelvis clip clip n...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92546</th>\n",
       "      <td>99992</td>\n",
       "      <td>2144-07-27</td>\n",
       "      <td>patient test information indication wall motio...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1116220</th>\n",
       "      <td>99991</td>\n",
       "      <td>2185-01-01</td>\n",
       "      <td>2185 1 1 3 02 pm heel axial lateral left clip...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49820</th>\n",
       "      <td>99985</td>\n",
       "      <td>2181-02-12</td>\n",
       "      <td>admission date 2181 1 27 discharge date 2181 2...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1674524</th>\n",
       "      <td>7</td>\n",
       "      <td>2121-05-25</td>\n",
       "      <td>npn septic evaluation infant was brought from ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>896528</th>\n",
       "      <td>6</td>\n",
       "      <td>2175-09-04</td>\n",
       "      <td>2175 9 4 7 38 am mra lower ext w w o contrast...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>813914</th>\n",
       "      <td>4</td>\n",
       "      <td>2191-06-23</td>\n",
       "      <td>2191 6 23 9 26 am mri abdomen w o w contrast ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>777408</th>\n",
       "      <td>3</td>\n",
       "      <td>2102-04-18</td>\n",
       "      <td>2102 4 18 1 22 pm art dup ext lo uni f u clip...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1671019</th>\n",
       "      <td>2</td>\n",
       "      <td>2138-07-17</td>\n",
       "      <td>nursing transfer note pt admitted to nicu for ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>38214 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         SUBJECT_ID        DATE  \\\n",
       "718850        99999  2118-01-01   \n",
       "1266755       99995  2147-03-16   \n",
       "92546         99992  2144-07-27   \n",
       "1116220       99991  2185-01-01   \n",
       "49820         99985  2181-02-12   \n",
       "...             ...         ...   \n",
       "1674524           7  2121-05-25   \n",
       "896528            6  2175-09-04   \n",
       "813914            4  2191-06-23   \n",
       "777408            3  2102-04-18   \n",
       "1671019           2  2138-07-17   \n",
       "\n",
       "                                                CLEAN_TEXT  \n",
       "718850   sicu hpi 63 f pod1 s p plif with respiratory d...  \n",
       "1266755   2147 3 16 11 24 am cta abd pelvis clip clip n...  \n",
       "92546    patient test information indication wall motio...  \n",
       "1116220   2185 1 1 3 02 pm heel axial lateral left clip...  \n",
       "49820    admission date 2181 1 27 discharge date 2181 2...  \n",
       "...                                                    ...  \n",
       "1674524  npn septic evaluation infant was brought from ...  \n",
       "896528    2175 9 4 7 38 am mra lower ext w w o contrast...  \n",
       "813914    2191 6 23 9 26 am mri abdomen w o w contrast ...  \n",
       "777408    2102 4 18 1 22 pm art dup ext lo uni f u clip...  \n",
       "1671019  nursing transfer note pt admitted to nicu for ...  \n",
       "\n",
       "[38214 rows x 3 columns]"
      ]
     },
     "execution_count": 92,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "last_note"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [],
   "source": [
    "stop_words = text.ENGLISH_STOP_WORDS\n",
    "vectorizer = TfidfVectorizer(max_features=200, stop_words=stop_words)\n",
    "tf_idf = vectorizer.fit_transform(last_note.CLEAN_TEXT)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols = [f\"TFIDF_{feat}\" for feat in vectorizer.get_feature_names()]\n",
    "tf_idf_feats = pd.DataFrame.sparse.from_spmatrix(tf_idf, columns=cols, index=last_note.SUBJECT_ID)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 268,
   "metadata": {},
   "outputs": [],
   "source": [
    "df, target = build_df_target(diag_final, meds_final, labs_final, feats_to_merge_left=[tf_idf_feats])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 269,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False    29631\n",
       "True      9280\n",
       "Name: target, dtype: int64"
      ]
     },
     "execution_count": 269,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "target.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 270,
   "metadata": {},
   "outputs": [],
   "source": [
    "cl = RandomForestClassifier(random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 271,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = train_and_predict(df, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 272,
   "metadata": {
    "collapsed": true,
    "jupyter": {
     "outputs_hidden": true
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.896\n",
      "\n",
      "LAB_51006_max       0.013008\n",
      "LAB_51006_mean      0.007803\n",
      "LAB_51277_mean      0.006911\n",
      "LAB_51249_min       0.006067\n",
      "LAB_51277_max       0.005876\n",
      "TFIDF_service       0.005007\n",
      "LAB_51277_min       0.005004\n",
      "LAB_50912_max       0.004958\n",
      "LAB_50931_max       0.004546\n",
      "LAB_50910_min       0.004249\n",
      "LAB_51006_min       0.004180\n",
      "LAB_50912_mean      0.004169\n",
      "LAB_51256_max       0.004167\n",
      "LAB_51275_mean      0.004086\n",
      "LAB_51274_max       0.003979\n",
      "TFIDF_discharge     0.003953\n",
      "TFIDF_date          0.003872\n",
      "LAB_51279_min       0.003855\n",
      "LAB_50970_max       0.003822\n",
      "LAB_50912_min       0.003798\n",
      "LAB_50863_mean      0.003730\n",
      "LAB_50863_max       0.003716\n",
      "LAB_51249_mean      0.003558\n",
      "LAB_51218_max       0.003512\n",
      "LAB_51218_mean      0.003482\n",
      "LAB_51222_min       0.003434\n",
      "TFIDF_diagnosis     0.003366\n",
      "TFIDF_admitting     0.003294\n",
      "LAB_50893_mean      0.003281\n",
      "LAB_51275_max       0.003266\n",
      "MED_027462_count    0.003257\n",
      "LAB_51222_mean      0.003238\n",
      "LAB_51274_min       0.003185\n",
      "TFIDF_mg            0.003175\n",
      "LAB_50931_min       0.003155\n",
      "LAB_50993_max       0.003133\n",
      "LAB_50971_max       0.003069\n",
      "LAB_51218_min       0.002889\n",
      "LAB_50893_max       0.002835\n",
      "LAB_50993_mean      0.002827\n",
      "LAB_50902_mean      0.002771\n",
      "TFIDF_showed        0.002733\n",
      "LAB_51237_max       0.002720\n",
      "LAB_51237_min       0.002707\n",
      "TFIDF_past          0.002664\n",
      "LAB_51274_mean      0.002659\n",
      "LAB_50970_mean      0.002595\n",
      "LAB_50960_mean      0.002594\n",
      "TFIDF_tablet        0.002571\n",
      "TFIDF_admission     0.002562\n",
      "LAB_50882_mean      0.002558\n",
      "ICD9_4280_count     0.002541\n",
      "LAB_51222_max       0.002535\n",
      "MED_nan_mean        0.002516\n",
      "LAB_51256_mean      0.002515\n",
      "LAB_50862_min       0.002504\n",
      "LAB_50893_min       0.002484\n",
      "LAB_51221_min       0.002484\n",
      "LAB_51265_mean      0.002469\n",
      "LAB_50910_mean      0.002434\n",
      "TFIDF_patient       0.002434\n",
      "TFIDF_sig           0.002422\n",
      "LAB_51301_min       0.002406\n",
      "TFIDF_course        0.002406\n",
      "LAB_51301_mean      0.002393\n",
      "LAB_50902_min       0.002392\n",
      "LAB_50983_mean      0.002375\n",
      "LAB_51250_mean      0.002341\n",
      "LAB_51003_mean      0.002306\n",
      "TFIDF_10            0.002301\n",
      "TFIDF_surgical      0.002257\n",
      "LAB_51265_max       0.002237\n",
      "TFIDF_care          0.002228\n",
      "LAB_51237_mean      0.002221\n",
      "LAB_50863_min       0.002220\n",
      "LAB_50910_max       0.002202\n",
      "TFIDF_unit          0.002186\n",
      "TFIDF_times         0.002185\n",
      "LAB_51265_min       0.002185\n",
      "LAB_50878_mean      0.002166\n",
      "LAB_51498_mean      0.002159\n",
      "LAB_51279_max       0.002155\n",
      "LAB_50862_max       0.002145\n",
      "MED_027462_mean     0.002127\n",
      "LAB_51254_max       0.002122\n",
      "LAB_51221_mean      0.002119\n",
      "LAB_51221_max       0.002113\n",
      "LAB_50931_mean      0.002110\n",
      "LAB_51244_min       0.002110\n",
      "LAB_50868_mean      0.002105\n",
      "LAB_51301_max       0.002102\n",
      "LAB_50953_max       0.002099\n",
      "LAB_51244_max       0.002095\n",
      "LAB_51275_min       0.002087\n",
      "LAB_50862_mean      0.002084\n",
      "LAB_51250_max       0.002062\n",
      "TFIDF_continued     0.002028\n",
      "LAB_50878_min       0.002016\n",
      "LAB_50971_mean      0.002015\n",
      "LAB_51248_mean      0.002010\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, cl, show_top_feats=100, col_names=df.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Better results, mostly from getting patient discharge information from notes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top 500 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 6319), transformed shape: (31165, 500)\n",
      "CPU times: user 44.5 s, sys: 3.78 s, total: 48.3 s\n",
      "Wall time: 49.7 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 237,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.894\n",
      "\n",
      "LAB_51006_max     0.020105\n",
      "LAB_51006_mean    0.018432\n",
      "LAB_51277_max     0.014042\n",
      "LAB_51277_mean    0.011792\n",
      "TFIDF_service     0.011593\n",
      "LAB_50912_max     0.011003\n",
      "LAB_51249_min     0.008346\n",
      "LAB_51277_min     0.007501\n",
      "TFIDF_date        0.007240\n",
      "LAB_51256_max     0.007001\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top features based on selection strategy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 238,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 6319), transformed shape: (31165, 979)\n",
      "CPU times: user 49.8 s, sys: 2.75 s, total: 52.6 s\n",
      "Wall time: 53.4 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df, target, cl, threshold=None, max_feats=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 239,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.900\n",
      "\n",
      "LAB_51006_max      0.017160\n",
      "LAB_51006_mean     0.014646\n",
      "LAB_51277_max      0.010655\n",
      "TFIDF_discharge    0.009754\n",
      "LAB_51277_mean     0.009102\n",
      "TFIDF_service      0.007165\n",
      "LAB_51249_min      0.007075\n",
      "LAB_51256_max      0.006914\n",
      "LAB_50912_max      0.006914\n",
      "LAB_51277_min      0.006708\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add transformer embeddings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "LAST_NOTE_EMBEDS_PATH = \"../data/embeddings/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>DATE</th>\n",
       "      <th>CLEAN_TEXT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>718850</th>\n",
       "      <td>99999</td>\n",
       "      <td>2118-01-01</td>\n",
       "      <td>sicu hpi 63 f pod1 s p plif with respiratory d...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1266755</th>\n",
       "      <td>99995</td>\n",
       "      <td>2147-03-16</td>\n",
       "      <td>2147 3 16 11 24 am cta abd pelvis clip clip n...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92546</th>\n",
       "      <td>99992</td>\n",
       "      <td>2144-07-27</td>\n",
       "      <td>patient test information indication wall motio...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1116220</th>\n",
       "      <td>99991</td>\n",
       "      <td>2185-01-01</td>\n",
       "      <td>2185 1 1 3 02 pm heel axial lateral left clip...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49820</th>\n",
       "      <td>99985</td>\n",
       "      <td>2181-02-12</td>\n",
       "      <td>admission date 2181 1 27 discharge date 2181 2...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1674524</th>\n",
       "      <td>7</td>\n",
       "      <td>2121-05-25</td>\n",
       "      <td>npn septic evaluation infant was brought from ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>896528</th>\n",
       "      <td>6</td>\n",
       "      <td>2175-09-04</td>\n",
       "      <td>2175 9 4 7 38 am mra lower ext w w o contrast...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>813914</th>\n",
       "      <td>4</td>\n",
       "      <td>2191-06-23</td>\n",
       "      <td>2191 6 23 9 26 am mri abdomen w o w contrast ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>777408</th>\n",
       "      <td>3</td>\n",
       "      <td>2102-04-18</td>\n",
       "      <td>2102 4 18 1 22 pm art dup ext lo uni f u clip...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1671019</th>\n",
       "      <td>2</td>\n",
       "      <td>2138-07-17</td>\n",
       "      <td>nursing transfer note pt admitted to nicu for ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>38214 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         SUBJECT_ID        DATE  \\\n",
       "718850        99999  2118-01-01   \n",
       "1266755       99995  2147-03-16   \n",
       "92546         99992  2144-07-27   \n",
       "1116220       99991  2185-01-01   \n",
       "49820         99985  2181-02-12   \n",
       "...             ...         ...   \n",
       "1674524           7  2121-05-25   \n",
       "896528            6  2175-09-04   \n",
       "813914            4  2191-06-23   \n",
       "777408            3  2102-04-18   \n",
       "1671019           2  2138-07-17   \n",
       "\n",
       "                                                CLEAN_TEXT  \n",
       "718850   sicu hpi 63 f pod1 s p plif with respiratory d...  \n",
       "1266755   2147 3 16 11 24 am cta abd pelvis clip clip n...  \n",
       "92546    patient test information indication wall motio...  \n",
       "1116220   2185 1 1 3 02 pm heel axial lateral left clip...  \n",
       "49820    admission date 2181 1 27 discharge date 2181 2...  \n",
       "...                                                    ...  \n",
       "1674524  npn septic evaluation infant was brought from ...  \n",
       "896528    2175 9 4 7 38 am mra lower ext w w o contrast...  \n",
       "813914    2191 6 23 9 26 am mri abdomen w o w contrast ...  \n",
       "777408    2102 4 18 1 22 pm art dup ext lo uni f u clip...  \n",
       "1671019  nursing transfer note pt admitted to nicu for ...  \n",
       "\n",
       "[38214 rows x 3 columns]"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "last_note"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "# prepare_text_for_tokenizer(last_note['TEXT'].iloc[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "last_note['TO_TOK'] = last_note.CLEAN_TEXT.map(prepare_text_for_tokenizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "last_note = last_note.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>DATE</th>\n",
       "      <th>CLEAN_TEXT</th>\n",
       "      <th>TO_TOK</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>99999</td>\n",
       "      <td>2118-01-01</td>\n",
       "      <td>sicu hpi 63 f pod1 s p plif with respiratory d...</td>\n",
       "      <td>sicu hpi 63 f pod1 s p plif with respiratory d...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>99995</td>\n",
       "      <td>2147-03-16</td>\n",
       "      <td>2147 3 16 11 24 am cta abd pelvis clip clip n...</td>\n",
       "      <td>2147 3 16 11 24 am cta abd pelvis clip clip n...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>99992</td>\n",
       "      <td>2144-07-27</td>\n",
       "      <td>patient test information indication wall motio...</td>\n",
       "      <td>patient test information indication wall motio...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>99991</td>\n",
       "      <td>2185-01-01</td>\n",
       "      <td>2185 1 1 3 02 pm heel axial lateral left clip...</td>\n",
       "      <td>2185 1 1 3 02 pm heel axial lateral left clip...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>99985</td>\n",
       "      <td>2181-02-12</td>\n",
       "      <td>admission date 2181 1 27 discharge date 2181 2...</td>\n",
       "      <td>admission date 2181 1 27 discharge date 2181 2...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   SUBJECT_ID        DATE                                         CLEAN_TEXT  \\\n",
       "0       99999  2118-01-01  sicu hpi 63 f pod1 s p plif with respiratory d...   \n",
       "1       99995  2147-03-16   2147 3 16 11 24 am cta abd pelvis clip clip n...   \n",
       "2       99992  2144-07-27  patient test information indication wall motio...   \n",
       "3       99991  2185-01-01   2185 1 1 3 02 pm heel axial lateral left clip...   \n",
       "4       99985  2181-02-12  admission date 2181 1 27 discharge date 2181 2...   \n",
       "\n",
       "                                              TO_TOK  \n",
       "0  sicu hpi 63 f pod1 s p plif with respiratory d...  \n",
       "1   2147 3 16 11 24 am cta abd pelvis clip clip n...  \n",
       "2  patient test information indication wall motio...  \n",
       "3   2185 1 1 3 02 pm heel axial lateral left clip...  \n",
       "4  admission date 2181 1 27 discharge date 2181 2...  "
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "last_note.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = AutoTokenizer.from_pretrained(\"deepset/covid_bert_base\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [],
   "source": [
    "config = AutoConfig.from_pretrained('deepset/covid_bert_base', output_hidden_states=True, output_attentions=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "model = AutoModelForMaskedLM.from_pretrained(\"deepset/covid_bert_base\", config=config)\n",
    "model_structure = model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_vector_for_text(text: str, model, tokenizer, max_length: Optional[int] = None) -> torch.Tensor:\n",
    "    \"\"\"This can be made faster.\"\"\"\n",
    "    encoding = tokenizer(text, \n",
    "                         add_special_tokens=True, \n",
    "                         truncation=True, \n",
    "                         padding=\"max_length\", \n",
    "                         return_attention_mask=True, \n",
    "                         return_tensors=\"pt\", \n",
    "                         max_length=max_length)\n",
    "    with torch.no_grad():\n",
    "        outputs = model(**encoding)\n",
    "        hs = outputs.hidden_states\n",
    "        token_embeddings = torch.stack(hs, dim=0)\n",
    "        token_embeddings = torch.squeeze(token_embeddings, dim=1)\n",
    "        token_embeddings = token_embeddings.permute(1,0,2)\n",
    "        token_vecs = hs[-2][0]\n",
    "        text_embedding = torch.mean(token_vecs, dim=0)\n",
    "        return text_embedding"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 115,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculate_and_store_embeds(path: str, tokenizer, model, df: pd.DataFrame, \n",
    "                               text_col: str, embed_id: str = 'SUBJECT_ID', max_length: Optional[int] = None) -> None:\n",
    "    \n",
    "    for row_num, row in tqdm(df.iterrows(), \n",
    "                             total=df.shape[0]):\n",
    "        text = row[text_col]\n",
    "        subj_id = row[embed_id]\n",
    "        embedding = get_vector_for_text(text, model, tokenizer, max_length=max_length)\n",
    "        torch.save(embedding, os.path.join(path, f\"{subj_id}.pt\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 116,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 3.23 s, sys: 253 ms, total: 3.48 s\n",
      "Wall time: 995 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sentence = get_vector_for_text(last_note.TO_TOK.iloc[0], model, tokenizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "a7685fbcf1e24a219143f2da279b3ecb",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/38214 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "calculate_and_store_embeds(LAST_NOTE_EMBEDS_PATH,\n",
    "                           tokenizer, model,\n",
    "                           last_note,\n",
    "                           text_col='TO_TOK')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_embeds_df(path: str) -> pd.DataFrame:\n",
    "    subj_ids = []\n",
    "    embeds = []\n",
    "    for file in tqdm(glob(path + '*')):\n",
    "        name = file.split('/')[-1]\n",
    "        subj_id = int(name.split('.')[0])\n",
    "        embedding = torch.load(file)\n",
    "        subj_ids.append(subj_id)\n",
    "        embeds.append(np.array(embedding))\n",
    "    embed_df = pd.DataFrame(embeds, index=subj_ids)\n",
    "    embed_df.columns = [f\"EMBED_{i}\" for i in embed_df.columns]\n",
    "    return embed_df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "c387e489165749deb6bd79a7f8c95d9a",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/38214 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "embed_df_cord = get_embeds_df(LAST_NOTE_EMBEDS_PATH)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>EMBED_0</th>\n",
       "      <th>EMBED_1</th>\n",
       "      <th>EMBED_2</th>\n",
       "      <th>EMBED_3</th>\n",
       "      <th>EMBED_4</th>\n",
       "      <th>EMBED_5</th>\n",
       "      <th>EMBED_6</th>\n",
       "      <th>EMBED_7</th>\n",
       "      <th>EMBED_8</th>\n",
       "      <th>EMBED_9</th>\n",
       "      <th>EMBED_10</th>\n",
       "      <th>EMBED_11</th>\n",
       "      <th>EMBED_12</th>\n",
       "      <th>EMBED_13</th>\n",
       "      <th>EMBED_14</th>\n",
       "      <th>EMBED_15</th>\n",
       "      <th>EMBED_16</th>\n",
       "      <th>EMBED_17</th>\n",
       "      <th>EMBED_18</th>\n",
       "      <th>EMBED_19</th>\n",
       "      <th>EMBED_20</th>\n",
       "      <th>EMBED_21</th>\n",
       "      <th>EMBED_22</th>\n",
       "      <th>EMBED_23</th>\n",
       "      <th>EMBED_24</th>\n",
       "      <th>EMBED_25</th>\n",
       "      <th>EMBED_26</th>\n",
       "      <th>EMBED_27</th>\n",
       "      <th>EMBED_28</th>\n",
       "      <th>EMBED_29</th>\n",
       "      <th>EMBED_30</th>\n",
       "      <th>EMBED_31</th>\n",
       "      <th>EMBED_32</th>\n",
       "      <th>EMBED_33</th>\n",
       "      <th>EMBED_34</th>\n",
       "      <th>EMBED_35</th>\n",
       "      <th>EMBED_36</th>\n",
       "      <th>EMBED_37</th>\n",
       "      <th>EMBED_38</th>\n",
       "      <th>EMBED_39</th>\n",
       "      <th>EMBED_40</th>\n",
       "      <th>EMBED_41</th>\n",
       "      <th>EMBED_42</th>\n",
       "      <th>EMBED_43</th>\n",
       "      <th>EMBED_44</th>\n",
       "      <th>EMBED_45</th>\n",
       "      <th>EMBED_46</th>\n",
       "      <th>EMBED_47</th>\n",
       "      <th>EMBED_48</th>\n",
       "      <th>EMBED_49</th>\n",
       "      <th>EMBED_50</th>\n",
       "      <th>EMBED_51</th>\n",
       "      <th>EMBED_52</th>\n",
       "      <th>EMBED_53</th>\n",
       "      <th>EMBED_54</th>\n",
       "      <th>EMBED_55</th>\n",
       "      <th>EMBED_56</th>\n",
       "      <th>EMBED_57</th>\n",
       "      <th>EMBED_58</th>\n",
       "      <th>EMBED_59</th>\n",
       "      <th>EMBED_60</th>\n",
       "      <th>EMBED_61</th>\n",
       "      <th>EMBED_62</th>\n",
       "      <th>EMBED_63</th>\n",
       "      <th>EMBED_64</th>\n",
       "      <th>EMBED_65</th>\n",
       "      <th>EMBED_66</th>\n",
       "      <th>EMBED_67</th>\n",
       "      <th>EMBED_68</th>\n",
       "      <th>EMBED_69</th>\n",
       "      <th>EMBED_70</th>\n",
       "      <th>EMBED_71</th>\n",
       "      <th>EMBED_72</th>\n",
       "      <th>EMBED_73</th>\n",
       "      <th>EMBED_74</th>\n",
       "      <th>...</th>\n",
       "      <th>EMBED_693</th>\n",
       "      <th>EMBED_694</th>\n",
       "      <th>EMBED_695</th>\n",
       "      <th>EMBED_696</th>\n",
       "      <th>EMBED_697</th>\n",
       "      <th>EMBED_698</th>\n",
       "      <th>EMBED_699</th>\n",
       "      <th>EMBED_700</th>\n",
       "      <th>EMBED_701</th>\n",
       "      <th>EMBED_702</th>\n",
       "      <th>EMBED_703</th>\n",
       "      <th>EMBED_704</th>\n",
       "      <th>EMBED_705</th>\n",
       "      <th>EMBED_706</th>\n",
       "      <th>EMBED_707</th>\n",
       "      <th>EMBED_708</th>\n",
       "      <th>EMBED_709</th>\n",
       "      <th>EMBED_710</th>\n",
       "      <th>EMBED_711</th>\n",
       "      <th>EMBED_712</th>\n",
       "      <th>EMBED_713</th>\n",
       "      <th>EMBED_714</th>\n",
       "      <th>EMBED_715</th>\n",
       "      <th>EMBED_716</th>\n",
       "      <th>EMBED_717</th>\n",
       "      <th>EMBED_718</th>\n",
       "      <th>EMBED_719</th>\n",
       "      <th>EMBED_720</th>\n",
       "      <th>EMBED_721</th>\n",
       "      <th>EMBED_722</th>\n",
       "      <th>EMBED_723</th>\n",
       "      <th>EMBED_724</th>\n",
       "      <th>EMBED_725</th>\n",
       "      <th>EMBED_726</th>\n",
       "      <th>EMBED_727</th>\n",
       "      <th>EMBED_728</th>\n",
       "      <th>EMBED_729</th>\n",
       "      <th>EMBED_730</th>\n",
       "      <th>EMBED_731</th>\n",
       "      <th>EMBED_732</th>\n",
       "      <th>EMBED_733</th>\n",
       "      <th>EMBED_734</th>\n",
       "      <th>EMBED_735</th>\n",
       "      <th>EMBED_736</th>\n",
       "      <th>EMBED_737</th>\n",
       "      <th>EMBED_738</th>\n",
       "      <th>EMBED_739</th>\n",
       "      <th>EMBED_740</th>\n",
       "      <th>EMBED_741</th>\n",
       "      <th>EMBED_742</th>\n",
       "      <th>EMBED_743</th>\n",
       "      <th>EMBED_744</th>\n",
       "      <th>EMBED_745</th>\n",
       "      <th>EMBED_746</th>\n",
       "      <th>EMBED_747</th>\n",
       "      <th>EMBED_748</th>\n",
       "      <th>EMBED_749</th>\n",
       "      <th>EMBED_750</th>\n",
       "      <th>EMBED_751</th>\n",
       "      <th>EMBED_752</th>\n",
       "      <th>EMBED_753</th>\n",
       "      <th>EMBED_754</th>\n",
       "      <th>EMBED_755</th>\n",
       "      <th>EMBED_756</th>\n",
       "      <th>EMBED_757</th>\n",
       "      <th>EMBED_758</th>\n",
       "      <th>EMBED_759</th>\n",
       "      <th>EMBED_760</th>\n",
       "      <th>EMBED_761</th>\n",
       "      <th>EMBED_762</th>\n",
       "      <th>EMBED_763</th>\n",
       "      <th>EMBED_764</th>\n",
       "      <th>EMBED_765</th>\n",
       "      <th>EMBED_766</th>\n",
       "      <th>EMBED_767</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>71722</th>\n",
       "      <td>-0.199597</td>\n",
       "      <td>0.128276</td>\n",
       "      <td>0.718085</td>\n",
       "      <td>-0.388701</td>\n",
       "      <td>-0.102329</td>\n",
       "      <td>0.167867</td>\n",
       "      <td>0.029645</td>\n",
       "      <td>-0.340557</td>\n",
       "      <td>0.107918</td>\n",
       "      <td>-0.422439</td>\n",
       "      <td>0.273899</td>\n",
       "      <td>-0.099894</td>\n",
       "      <td>-0.046134</td>\n",
       "      <td>-0.381045</td>\n",
       "      <td>0.176543</td>\n",
       "      <td>0.717357</td>\n",
       "      <td>-0.057874</td>\n",
       "      <td>0.002464</td>\n",
       "      <td>-0.113914</td>\n",
       "      <td>0.183621</td>\n",
       "      <td>0.630027</td>\n",
       "      <td>-0.357316</td>\n",
       "      <td>0.291080</td>\n",
       "      <td>0.354838</td>\n",
       "      <td>-0.001738</td>\n",
       "      <td>0.175875</td>\n",
       "      <td>-0.071201</td>\n",
       "      <td>-0.069151</td>\n",
       "      <td>0.142517</td>\n",
       "      <td>-0.114328</td>\n",
       "      <td>0.233246</td>\n",
       "      <td>0.152829</td>\n",
       "      <td>0.109970</td>\n",
       "      <td>-0.527347</td>\n",
       "      <td>0.316943</td>\n",
       "      <td>0.050168</td>\n",
       "      <td>-0.174813</td>\n",
       "      <td>0.018598</td>\n",
       "      <td>-0.063050</td>\n",
       "      <td>-0.246874</td>\n",
       "      <td>-0.049857</td>\n",
       "      <td>-0.081755</td>\n",
       "      <td>0.008914</td>\n",
       "      <td>-0.299153</td>\n",
       "      <td>-0.541124</td>\n",
       "      <td>0.022232</td>\n",
       "      <td>0.144702</td>\n",
       "      <td>0.223001</td>\n",
       "      <td>-0.427878</td>\n",
       "      <td>0.175729</td>\n",
       "      <td>-0.437079</td>\n",
       "      <td>-0.455686</td>\n",
       "      <td>0.011744</td>\n",
       "      <td>-0.188377</td>\n",
       "      <td>-0.098222</td>\n",
       "      <td>0.186742</td>\n",
       "      <td>0.007582</td>\n",
       "      <td>-0.007565</td>\n",
       "      <td>-0.099802</td>\n",
       "      <td>-0.263568</td>\n",
       "      <td>0.166673</td>\n",
       "      <td>0.264614</td>\n",
       "      <td>-0.297953</td>\n",
       "      <td>0.126591</td>\n",
       "      <td>-0.294990</td>\n",
       "      <td>0.034109</td>\n",
       "      <td>-0.291699</td>\n",
       "      <td>0.186357</td>\n",
       "      <td>-0.169276</td>\n",
       "      <td>0.559821</td>\n",
       "      <td>-0.136239</td>\n",
       "      <td>0.065150</td>\n",
       "      <td>0.435436</td>\n",
       "      <td>0.272192</td>\n",
       "      <td>0.122566</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.074757</td>\n",
       "      <td>-0.100806</td>\n",
       "      <td>-0.128544</td>\n",
       "      <td>0.023274</td>\n",
       "      <td>-0.284944</td>\n",
       "      <td>-0.659821</td>\n",
       "      <td>0.027020</td>\n",
       "      <td>0.525081</td>\n",
       "      <td>-0.761540</td>\n",
       "      <td>0.068248</td>\n",
       "      <td>-0.071535</td>\n",
       "      <td>-0.207143</td>\n",
       "      <td>-0.391398</td>\n",
       "      <td>0.365011</td>\n",
       "      <td>-0.041412</td>\n",
       "      <td>-0.347234</td>\n",
       "      <td>0.137552</td>\n",
       "      <td>-0.297956</td>\n",
       "      <td>0.055337</td>\n",
       "      <td>0.616171</td>\n",
       "      <td>-0.387249</td>\n",
       "      <td>0.127927</td>\n",
       "      <td>0.400709</td>\n",
       "      <td>-0.261731</td>\n",
       "      <td>-0.044241</td>\n",
       "      <td>-0.150619</td>\n",
       "      <td>0.193708</td>\n",
       "      <td>0.128232</td>\n",
       "      <td>0.226463</td>\n",
       "      <td>-0.086620</td>\n",
       "      <td>-0.095651</td>\n",
       "      <td>-0.471604</td>\n",
       "      <td>-0.369074</td>\n",
       "      <td>-0.236608</td>\n",
       "      <td>-0.208466</td>\n",
       "      <td>-0.676335</td>\n",
       "      <td>-0.396741</td>\n",
       "      <td>0.048930</td>\n",
       "      <td>-0.177669</td>\n",
       "      <td>-0.209846</td>\n",
       "      <td>-0.297194</td>\n",
       "      <td>-0.242546</td>\n",
       "      <td>0.043388</td>\n",
       "      <td>0.051675</td>\n",
       "      <td>0.110699</td>\n",
       "      <td>-0.286888</td>\n",
       "      <td>0.320427</td>\n",
       "      <td>-0.144408</td>\n",
       "      <td>0.121379</td>\n",
       "      <td>0.426598</td>\n",
       "      <td>-0.109743</td>\n",
       "      <td>-0.043318</td>\n",
       "      <td>0.287685</td>\n",
       "      <td>0.194216</td>\n",
       "      <td>0.350567</td>\n",
       "      <td>0.007508</td>\n",
       "      <td>-0.344494</td>\n",
       "      <td>0.273991</td>\n",
       "      <td>0.323298</td>\n",
       "      <td>0.683664</td>\n",
       "      <td>0.023797</td>\n",
       "      <td>-0.576798</td>\n",
       "      <td>0.000249</td>\n",
       "      <td>-0.021221</td>\n",
       "      <td>-0.588558</td>\n",
       "      <td>-0.190467</td>\n",
       "      <td>0.273815</td>\n",
       "      <td>-0.097742</td>\n",
       "      <td>-0.292752</td>\n",
       "      <td>-0.022243</td>\n",
       "      <td>-0.217447</td>\n",
       "      <td>-0.979703</td>\n",
       "      <td>0.294535</td>\n",
       "      <td>-0.416701</td>\n",
       "      <td>0.165263</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>82412</th>\n",
       "      <td>-0.468055</td>\n",
       "      <td>0.430632</td>\n",
       "      <td>0.723761</td>\n",
       "      <td>-0.488057</td>\n",
       "      <td>0.030809</td>\n",
       "      <td>0.204566</td>\n",
       "      <td>0.286588</td>\n",
       "      <td>-0.089070</td>\n",
       "      <td>-0.068675</td>\n",
       "      <td>-0.036630</td>\n",
       "      <td>0.107773</td>\n",
       "      <td>-0.220798</td>\n",
       "      <td>-0.264219</td>\n",
       "      <td>-0.279889</td>\n",
       "      <td>-0.175305</td>\n",
       "      <td>0.519388</td>\n",
       "      <td>-0.052233</td>\n",
       "      <td>-0.118592</td>\n",
       "      <td>-0.150916</td>\n",
       "      <td>0.275564</td>\n",
       "      <td>0.715271</td>\n",
       "      <td>-0.157016</td>\n",
       "      <td>0.192690</td>\n",
       "      <td>0.159444</td>\n",
       "      <td>0.211376</td>\n",
       "      <td>0.152046</td>\n",
       "      <td>0.033970</td>\n",
       "      <td>-0.332181</td>\n",
       "      <td>0.054970</td>\n",
       "      <td>0.003761</td>\n",
       "      <td>0.155690</td>\n",
       "      <td>0.253198</td>\n",
       "      <td>0.376423</td>\n",
       "      <td>-0.429835</td>\n",
       "      <td>0.224837</td>\n",
       "      <td>-0.067574</td>\n",
       "      <td>-0.071792</td>\n",
       "      <td>-0.275832</td>\n",
       "      <td>0.122419</td>\n",
       "      <td>-0.173609</td>\n",
       "      <td>-0.457992</td>\n",
       "      <td>-0.161361</td>\n",
       "      <td>-0.088098</td>\n",
       "      <td>-0.041622</td>\n",
       "      <td>-0.774563</td>\n",
       "      <td>0.168907</td>\n",
       "      <td>0.237651</td>\n",
       "      <td>0.141014</td>\n",
       "      <td>-0.585243</td>\n",
       "      <td>-0.089969</td>\n",
       "      <td>-0.441000</td>\n",
       "      <td>-0.412005</td>\n",
       "      <td>0.238198</td>\n",
       "      <td>-0.322015</td>\n",
       "      <td>-0.049432</td>\n",
       "      <td>0.081350</td>\n",
       "      <td>-0.264637</td>\n",
       "      <td>-0.287227</td>\n",
       "      <td>-0.276944</td>\n",
       "      <td>-0.359161</td>\n",
       "      <td>0.062582</td>\n",
       "      <td>0.092121</td>\n",
       "      <td>-0.136310</td>\n",
       "      <td>0.139558</td>\n",
       "      <td>-0.323911</td>\n",
       "      <td>-0.021034</td>\n",
       "      <td>-0.185649</td>\n",
       "      <td>0.255453</td>\n",
       "      <td>-0.063558</td>\n",
       "      <td>0.737593</td>\n",
       "      <td>-0.062536</td>\n",
       "      <td>-0.007713</td>\n",
       "      <td>0.547863</td>\n",
       "      <td>0.469338</td>\n",
       "      <td>0.166559</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.111920</td>\n",
       "      <td>-0.257764</td>\n",
       "      <td>-0.061592</td>\n",
       "      <td>0.127675</td>\n",
       "      <td>-0.116015</td>\n",
       "      <td>-0.591214</td>\n",
       "      <td>-0.203812</td>\n",
       "      <td>0.636429</td>\n",
       "      <td>-0.498779</td>\n",
       "      <td>-0.087474</td>\n",
       "      <td>-0.318900</td>\n",
       "      <td>-0.270757</td>\n",
       "      <td>-0.250299</td>\n",
       "      <td>0.688977</td>\n",
       "      <td>-0.062132</td>\n",
       "      <td>-0.351459</td>\n",
       "      <td>-0.502622</td>\n",
       "      <td>-0.393250</td>\n",
       "      <td>-0.065938</td>\n",
       "      <td>0.630741</td>\n",
       "      <td>-0.412833</td>\n",
       "      <td>-0.060410</td>\n",
       "      <td>0.451292</td>\n",
       "      <td>-0.372684</td>\n",
       "      <td>-0.172211</td>\n",
       "      <td>-0.214286</td>\n",
       "      <td>0.407914</td>\n",
       "      <td>0.042372</td>\n",
       "      <td>0.319015</td>\n",
       "      <td>-0.167069</td>\n",
       "      <td>0.006642</td>\n",
       "      <td>-0.352651</td>\n",
       "      <td>-0.094960</td>\n",
       "      <td>-0.064348</td>\n",
       "      <td>-0.154181</td>\n",
       "      <td>-0.592284</td>\n",
       "      <td>-0.539266</td>\n",
       "      <td>0.355212</td>\n",
       "      <td>0.099570</td>\n",
       "      <td>-0.217730</td>\n",
       "      <td>-0.284940</td>\n",
       "      <td>-0.216200</td>\n",
       "      <td>0.059145</td>\n",
       "      <td>0.097196</td>\n",
       "      <td>-0.305455</td>\n",
       "      <td>-0.279240</td>\n",
       "      <td>0.402879</td>\n",
       "      <td>0.214084</td>\n",
       "      <td>0.278291</td>\n",
       "      <td>0.127696</td>\n",
       "      <td>0.136121</td>\n",
       "      <td>-0.066565</td>\n",
       "      <td>0.282321</td>\n",
       "      <td>0.198715</td>\n",
       "      <td>0.477652</td>\n",
       "      <td>0.217438</td>\n",
       "      <td>-0.242581</td>\n",
       "      <td>0.337069</td>\n",
       "      <td>0.310483</td>\n",
       "      <td>0.532939</td>\n",
       "      <td>-0.021243</td>\n",
       "      <td>-0.505779</td>\n",
       "      <td>-0.126655</td>\n",
       "      <td>-0.135804</td>\n",
       "      <td>-0.814749</td>\n",
       "      <td>-0.441541</td>\n",
       "      <td>0.225606</td>\n",
       "      <td>-0.030038</td>\n",
       "      <td>-0.071902</td>\n",
       "      <td>0.058272</td>\n",
       "      <td>-0.225572</td>\n",
       "      <td>-0.901695</td>\n",
       "      <td>0.082120</td>\n",
       "      <td>-0.124130</td>\n",
       "      <td>-0.104794</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5114</th>\n",
       "      <td>-0.487482</td>\n",
       "      <td>0.475799</td>\n",
       "      <td>0.390588</td>\n",
       "      <td>-0.531058</td>\n",
       "      <td>0.287604</td>\n",
       "      <td>0.390556</td>\n",
       "      <td>0.258551</td>\n",
       "      <td>0.264478</td>\n",
       "      <td>0.159602</td>\n",
       "      <td>-0.254662</td>\n",
       "      <td>-0.011556</td>\n",
       "      <td>-0.112627</td>\n",
       "      <td>-0.166722</td>\n",
       "      <td>-0.191362</td>\n",
       "      <td>-0.092799</td>\n",
       "      <td>0.680919</td>\n",
       "      <td>0.216972</td>\n",
       "      <td>-0.148479</td>\n",
       "      <td>-0.508639</td>\n",
       "      <td>0.453912</td>\n",
       "      <td>1.507544</td>\n",
       "      <td>0.010893</td>\n",
       "      <td>0.603871</td>\n",
       "      <td>0.415800</td>\n",
       "      <td>0.221951</td>\n",
       "      <td>0.363705</td>\n",
       "      <td>-0.236890</td>\n",
       "      <td>-0.648948</td>\n",
       "      <td>0.115380</td>\n",
       "      <td>0.434274</td>\n",
       "      <td>0.357161</td>\n",
       "      <td>0.041595</td>\n",
       "      <td>0.505122</td>\n",
       "      <td>-0.471156</td>\n",
       "      <td>0.900400</td>\n",
       "      <td>-0.452694</td>\n",
       "      <td>-0.166682</td>\n",
       "      <td>-0.275302</td>\n",
       "      <td>-0.109170</td>\n",
       "      <td>0.014999</td>\n",
       "      <td>-0.648206</td>\n",
       "      <td>-0.278187</td>\n",
       "      <td>-0.153674</td>\n",
       "      <td>-0.036007</td>\n",
       "      <td>-0.217740</td>\n",
       "      <td>-0.103751</td>\n",
       "      <td>0.473294</td>\n",
       "      <td>-0.017797</td>\n",
       "      <td>-0.383454</td>\n",
       "      <td>0.001974</td>\n",
       "      <td>-0.505449</td>\n",
       "      <td>0.087838</td>\n",
       "      <td>0.134639</td>\n",
       "      <td>-0.086668</td>\n",
       "      <td>0.364900</td>\n",
       "      <td>0.125638</td>\n",
       "      <td>-0.328315</td>\n",
       "      <td>-0.014870</td>\n",
       "      <td>-0.419891</td>\n",
       "      <td>-0.190609</td>\n",
       "      <td>-0.222180</td>\n",
       "      <td>0.154361</td>\n",
       "      <td>0.249562</td>\n",
       "      <td>0.358595</td>\n",
       "      <td>-0.260617</td>\n",
       "      <td>0.187446</td>\n",
       "      <td>-0.156096</td>\n",
       "      <td>0.264043</td>\n",
       "      <td>-0.625066</td>\n",
       "      <td>0.065797</td>\n",
       "      <td>0.065141</td>\n",
       "      <td>-0.072569</td>\n",
       "      <td>0.474023</td>\n",
       "      <td>0.146674</td>\n",
       "      <td>0.593366</td>\n",
       "      <td>...</td>\n",
       "      <td>0.167698</td>\n",
       "      <td>-0.485078</td>\n",
       "      <td>0.149991</td>\n",
       "      <td>-0.262486</td>\n",
       "      <td>-0.443924</td>\n",
       "      <td>-0.202830</td>\n",
       "      <td>-0.599570</td>\n",
       "      <td>0.992166</td>\n",
       "      <td>-0.565094</td>\n",
       "      <td>-0.437035</td>\n",
       "      <td>0.200376</td>\n",
       "      <td>0.090037</td>\n",
       "      <td>-0.136571</td>\n",
       "      <td>1.011193</td>\n",
       "      <td>-0.069717</td>\n",
       "      <td>-0.165624</td>\n",
       "      <td>-0.296528</td>\n",
       "      <td>-0.951483</td>\n",
       "      <td>0.021219</td>\n",
       "      <td>0.649837</td>\n",
       "      <td>-0.118610</td>\n",
       "      <td>-0.220282</td>\n",
       "      <td>0.632864</td>\n",
       "      <td>-0.539973</td>\n",
       "      <td>-0.127614</td>\n",
       "      <td>-0.196914</td>\n",
       "      <td>-0.287379</td>\n",
       "      <td>0.224510</td>\n",
       "      <td>-0.198480</td>\n",
       "      <td>-0.553790</td>\n",
       "      <td>0.165419</td>\n",
       "      <td>-0.146234</td>\n",
       "      <td>-0.197197</td>\n",
       "      <td>-0.251285</td>\n",
       "      <td>-0.137213</td>\n",
       "      <td>-0.186979</td>\n",
       "      <td>-0.730872</td>\n",
       "      <td>0.166875</td>\n",
       "      <td>0.242221</td>\n",
       "      <td>0.085243</td>\n",
       "      <td>-0.731259</td>\n",
       "      <td>0.172389</td>\n",
       "      <td>-0.324964</td>\n",
       "      <td>-0.262773</td>\n",
       "      <td>-0.214583</td>\n",
       "      <td>-0.027991</td>\n",
       "      <td>0.414697</td>\n",
       "      <td>0.245987</td>\n",
       "      <td>0.057638</td>\n",
       "      <td>0.437427</td>\n",
       "      <td>0.161853</td>\n",
       "      <td>0.030441</td>\n",
       "      <td>0.244832</td>\n",
       "      <td>0.240373</td>\n",
       "      <td>0.684028</td>\n",
       "      <td>0.251944</td>\n",
       "      <td>0.229340</td>\n",
       "      <td>0.176840</td>\n",
       "      <td>0.306029</td>\n",
       "      <td>-0.189102</td>\n",
       "      <td>0.068314</td>\n",
       "      <td>-0.267470</td>\n",
       "      <td>-0.357423</td>\n",
       "      <td>-0.671371</td>\n",
       "      <td>-0.518839</td>\n",
       "      <td>-0.506295</td>\n",
       "      <td>0.155106</td>\n",
       "      <td>-0.004346</td>\n",
       "      <td>0.238540</td>\n",
       "      <td>0.260278</td>\n",
       "      <td>-0.108356</td>\n",
       "      <td>-1.002016</td>\n",
       "      <td>0.154130</td>\n",
       "      <td>0.441950</td>\n",
       "      <td>-0.105442</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17137</th>\n",
       "      <td>-0.291488</td>\n",
       "      <td>0.482612</td>\n",
       "      <td>0.659554</td>\n",
       "      <td>-0.425932</td>\n",
       "      <td>0.190169</td>\n",
       "      <td>0.185168</td>\n",
       "      <td>0.132073</td>\n",
       "      <td>-0.139042</td>\n",
       "      <td>0.050121</td>\n",
       "      <td>-0.451831</td>\n",
       "      <td>0.057091</td>\n",
       "      <td>-0.051556</td>\n",
       "      <td>-0.168111</td>\n",
       "      <td>-0.421862</td>\n",
       "      <td>-0.007254</td>\n",
       "      <td>0.543765</td>\n",
       "      <td>-0.094489</td>\n",
       "      <td>-0.185421</td>\n",
       "      <td>-0.325657</td>\n",
       "      <td>0.197056</td>\n",
       "      <td>0.800147</td>\n",
       "      <td>-0.115078</td>\n",
       "      <td>0.351718</td>\n",
       "      <td>0.373786</td>\n",
       "      <td>0.168323</td>\n",
       "      <td>0.222242</td>\n",
       "      <td>-0.088686</td>\n",
       "      <td>-0.163844</td>\n",
       "      <td>0.183858</td>\n",
       "      <td>0.086042</td>\n",
       "      <td>0.087384</td>\n",
       "      <td>0.203285</td>\n",
       "      <td>0.382363</td>\n",
       "      <td>-0.635049</td>\n",
       "      <td>0.420093</td>\n",
       "      <td>-0.106260</td>\n",
       "      <td>-0.208328</td>\n",
       "      <td>-0.035605</td>\n",
       "      <td>0.197733</td>\n",
       "      <td>-0.205399</td>\n",
       "      <td>-0.301760</td>\n",
       "      <td>-0.278275</td>\n",
       "      <td>-0.378255</td>\n",
       "      <td>-0.034065</td>\n",
       "      <td>-0.539690</td>\n",
       "      <td>-0.064060</td>\n",
       "      <td>0.140130</td>\n",
       "      <td>-0.144546</td>\n",
       "      <td>-0.702111</td>\n",
       "      <td>0.107039</td>\n",
       "      <td>-0.250343</td>\n",
       "      <td>-0.245399</td>\n",
       "      <td>0.247315</td>\n",
       "      <td>-0.089495</td>\n",
       "      <td>0.025982</td>\n",
       "      <td>0.262099</td>\n",
       "      <td>-0.102170</td>\n",
       "      <td>-0.187821</td>\n",
       "      <td>-0.167515</td>\n",
       "      <td>-0.331180</td>\n",
       "      <td>-0.017649</td>\n",
       "      <td>0.010149</td>\n",
       "      <td>-0.069578</td>\n",
       "      <td>0.314224</td>\n",
       "      <td>-0.188463</td>\n",
       "      <td>-0.055250</td>\n",
       "      <td>-0.186396</td>\n",
       "      <td>0.497756</td>\n",
       "      <td>-0.265870</td>\n",
       "      <td>0.462227</td>\n",
       "      <td>0.008109</td>\n",
       "      <td>-0.003724</td>\n",
       "      <td>0.591356</td>\n",
       "      <td>0.405798</td>\n",
       "      <td>0.459157</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.184273</td>\n",
       "      <td>-0.105833</td>\n",
       "      <td>0.005536</td>\n",
       "      <td>0.116708</td>\n",
       "      <td>-0.101085</td>\n",
       "      <td>-0.498121</td>\n",
       "      <td>0.005574</td>\n",
       "      <td>0.775227</td>\n",
       "      <td>-0.391825</td>\n",
       "      <td>0.032070</td>\n",
       "      <td>-0.277728</td>\n",
       "      <td>-0.439826</td>\n",
       "      <td>-0.271632</td>\n",
       "      <td>0.670824</td>\n",
       "      <td>-0.046978</td>\n",
       "      <td>-0.446914</td>\n",
       "      <td>-0.233220</td>\n",
       "      <td>-0.424722</td>\n",
       "      <td>0.022346</td>\n",
       "      <td>0.634328</td>\n",
       "      <td>-0.181058</td>\n",
       "      <td>0.138583</td>\n",
       "      <td>0.563455</td>\n",
       "      <td>-0.497472</td>\n",
       "      <td>-0.095692</td>\n",
       "      <td>-0.403126</td>\n",
       "      <td>0.205138</td>\n",
       "      <td>0.078813</td>\n",
       "      <td>0.323774</td>\n",
       "      <td>-0.172926</td>\n",
       "      <td>-0.181271</td>\n",
       "      <td>-0.288377</td>\n",
       "      <td>-0.299346</td>\n",
       "      <td>-0.488301</td>\n",
       "      <td>-0.061735</td>\n",
       "      <td>-0.507592</td>\n",
       "      <td>-0.529043</td>\n",
       "      <td>0.189368</td>\n",
       "      <td>-0.014910</td>\n",
       "      <td>-0.092923</td>\n",
       "      <td>-0.169689</td>\n",
       "      <td>-0.115287</td>\n",
       "      <td>0.007615</td>\n",
       "      <td>0.133118</td>\n",
       "      <td>-0.055452</td>\n",
       "      <td>-0.275068</td>\n",
       "      <td>0.560709</td>\n",
       "      <td>0.234336</td>\n",
       "      <td>-0.057133</td>\n",
       "      <td>0.114402</td>\n",
       "      <td>-0.006842</td>\n",
       "      <td>-0.142725</td>\n",
       "      <td>0.400997</td>\n",
       "      <td>0.196825</td>\n",
       "      <td>0.314641</td>\n",
       "      <td>0.263499</td>\n",
       "      <td>-0.360935</td>\n",
       "      <td>0.245796</td>\n",
       "      <td>0.327293</td>\n",
       "      <td>0.540557</td>\n",
       "      <td>-0.048752</td>\n",
       "      <td>-0.483916</td>\n",
       "      <td>-0.162530</td>\n",
       "      <td>0.071318</td>\n",
       "      <td>-0.695274</td>\n",
       "      <td>-0.291078</td>\n",
       "      <td>0.106476</td>\n",
       "      <td>0.135865</td>\n",
       "      <td>0.052395</td>\n",
       "      <td>0.025286</td>\n",
       "      <td>-0.130536</td>\n",
       "      <td>-1.048716</td>\n",
       "      <td>0.207678</td>\n",
       "      <td>-0.290715</td>\n",
       "      <td>-0.235323</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98280</th>\n",
       "      <td>0.051681</td>\n",
       "      <td>0.513536</td>\n",
       "      <td>0.803679</td>\n",
       "      <td>-0.444162</td>\n",
       "      <td>-0.099338</td>\n",
       "      <td>0.146255</td>\n",
       "      <td>-0.049577</td>\n",
       "      <td>-0.195717</td>\n",
       "      <td>-0.062002</td>\n",
       "      <td>-0.244867</td>\n",
       "      <td>-0.149850</td>\n",
       "      <td>-0.397408</td>\n",
       "      <td>0.034174</td>\n",
       "      <td>-0.095874</td>\n",
       "      <td>0.178256</td>\n",
       "      <td>0.564976</td>\n",
       "      <td>-0.289335</td>\n",
       "      <td>-0.127779</td>\n",
       "      <td>0.071391</td>\n",
       "      <td>0.310104</td>\n",
       "      <td>0.547856</td>\n",
       "      <td>0.163881</td>\n",
       "      <td>0.295819</td>\n",
       "      <td>0.224137</td>\n",
       "      <td>-0.278696</td>\n",
       "      <td>0.081467</td>\n",
       "      <td>-0.283003</td>\n",
       "      <td>-0.263821</td>\n",
       "      <td>0.184960</td>\n",
       "      <td>0.266168</td>\n",
       "      <td>0.057932</td>\n",
       "      <td>-0.132066</td>\n",
       "      <td>0.259967</td>\n",
       "      <td>-0.287199</td>\n",
       "      <td>0.383855</td>\n",
       "      <td>-0.101945</td>\n",
       "      <td>-0.196355</td>\n",
       "      <td>0.088330</td>\n",
       "      <td>0.207922</td>\n",
       "      <td>-0.158872</td>\n",
       "      <td>-0.079107</td>\n",
       "      <td>-0.076817</td>\n",
       "      <td>-0.413078</td>\n",
       "      <td>0.060489</td>\n",
       "      <td>-0.457690</td>\n",
       "      <td>0.310161</td>\n",
       "      <td>-0.064966</td>\n",
       "      <td>0.035300</td>\n",
       "      <td>-0.506857</td>\n",
       "      <td>-0.017609</td>\n",
       "      <td>-0.121277</td>\n",
       "      <td>-0.549041</td>\n",
       "      <td>-0.130149</td>\n",
       "      <td>-0.372267</td>\n",
       "      <td>-0.348622</td>\n",
       "      <td>0.538688</td>\n",
       "      <td>-0.004939</td>\n",
       "      <td>-0.150087</td>\n",
       "      <td>-0.305914</td>\n",
       "      <td>-0.177322</td>\n",
       "      <td>0.138986</td>\n",
       "      <td>-0.098812</td>\n",
       "      <td>0.206911</td>\n",
       "      <td>0.385512</td>\n",
       "      <td>-0.516083</td>\n",
       "      <td>-0.156244</td>\n",
       "      <td>-0.157140</td>\n",
       "      <td>0.357336</td>\n",
       "      <td>-0.147980</td>\n",
       "      <td>0.238990</td>\n",
       "      <td>-0.167907</td>\n",
       "      <td>0.039524</td>\n",
       "      <td>0.636579</td>\n",
       "      <td>0.384875</td>\n",
       "      <td>0.457279</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.138446</td>\n",
       "      <td>-0.120921</td>\n",
       "      <td>-0.183172</td>\n",
       "      <td>0.215548</td>\n",
       "      <td>0.123629</td>\n",
       "      <td>-0.479293</td>\n",
       "      <td>0.029013</td>\n",
       "      <td>0.311060</td>\n",
       "      <td>-0.414726</td>\n",
       "      <td>0.218956</td>\n",
       "      <td>-0.479729</td>\n",
       "      <td>-0.387204</td>\n",
       "      <td>-0.573244</td>\n",
       "      <td>0.474536</td>\n",
       "      <td>-0.132899</td>\n",
       "      <td>-0.246493</td>\n",
       "      <td>0.091711</td>\n",
       "      <td>-0.274285</td>\n",
       "      <td>-0.162845</td>\n",
       "      <td>0.533131</td>\n",
       "      <td>-0.743020</td>\n",
       "      <td>0.276702</td>\n",
       "      <td>0.706932</td>\n",
       "      <td>-0.541503</td>\n",
       "      <td>0.082545</td>\n",
       "      <td>-0.194933</td>\n",
       "      <td>0.461641</td>\n",
       "      <td>0.093110</td>\n",
       "      <td>0.463067</td>\n",
       "      <td>0.137677</td>\n",
       "      <td>-0.174352</td>\n",
       "      <td>-0.150621</td>\n",
       "      <td>-0.439288</td>\n",
       "      <td>-0.506649</td>\n",
       "      <td>-0.567832</td>\n",
       "      <td>-0.723359</td>\n",
       "      <td>-0.870768</td>\n",
       "      <td>-0.040746</td>\n",
       "      <td>0.053004</td>\n",
       "      <td>-0.174838</td>\n",
       "      <td>0.270660</td>\n",
       "      <td>-0.299528</td>\n",
       "      <td>-0.017289</td>\n",
       "      <td>0.525816</td>\n",
       "      <td>-0.245545</td>\n",
       "      <td>0.095457</td>\n",
       "      <td>0.530235</td>\n",
       "      <td>-0.098032</td>\n",
       "      <td>-0.017754</td>\n",
       "      <td>0.219984</td>\n",
       "      <td>0.148890</td>\n",
       "      <td>-0.016947</td>\n",
       "      <td>0.300868</td>\n",
       "      <td>0.194767</td>\n",
       "      <td>0.113305</td>\n",
       "      <td>0.148186</td>\n",
       "      <td>-0.247089</td>\n",
       "      <td>0.237505</td>\n",
       "      <td>0.325652</td>\n",
       "      <td>0.214886</td>\n",
       "      <td>-0.149150</td>\n",
       "      <td>-0.274708</td>\n",
       "      <td>0.031862</td>\n",
       "      <td>-0.309526</td>\n",
       "      <td>-0.674247</td>\n",
       "      <td>-0.339018</td>\n",
       "      <td>0.068102</td>\n",
       "      <td>0.615809</td>\n",
       "      <td>-0.321040</td>\n",
       "      <td>0.283644</td>\n",
       "      <td>-0.118222</td>\n",
       "      <td>-1.054533</td>\n",
       "      <td>-0.016321</td>\n",
       "      <td>-0.325125</td>\n",
       "      <td>0.209312</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 768 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        EMBED_0   EMBED_1   EMBED_2   EMBED_3   EMBED_4   EMBED_5   EMBED_6  \\\n",
       "71722 -0.199597  0.128276  0.718085 -0.388701 -0.102329  0.167867  0.029645   \n",
       "82412 -0.468055  0.430632  0.723761 -0.488057  0.030809  0.204566  0.286588   \n",
       "5114  -0.487482  0.475799  0.390588 -0.531058  0.287604  0.390556  0.258551   \n",
       "17137 -0.291488  0.482612  0.659554 -0.425932  0.190169  0.185168  0.132073   \n",
       "98280  0.051681  0.513536  0.803679 -0.444162 -0.099338  0.146255 -0.049577   \n",
       "\n",
       "        EMBED_7   EMBED_8   EMBED_9  EMBED_10  EMBED_11  EMBED_12  EMBED_13  \\\n",
       "71722 -0.340557  0.107918 -0.422439  0.273899 -0.099894 -0.046134 -0.381045   \n",
       "82412 -0.089070 -0.068675 -0.036630  0.107773 -0.220798 -0.264219 -0.279889   \n",
       "5114   0.264478  0.159602 -0.254662 -0.011556 -0.112627 -0.166722 -0.191362   \n",
       "17137 -0.139042  0.050121 -0.451831  0.057091 -0.051556 -0.168111 -0.421862   \n",
       "98280 -0.195717 -0.062002 -0.244867 -0.149850 -0.397408  0.034174 -0.095874   \n",
       "\n",
       "       EMBED_14  EMBED_15  EMBED_16  EMBED_17  EMBED_18  EMBED_19  EMBED_20  \\\n",
       "71722  0.176543  0.717357 -0.057874  0.002464 -0.113914  0.183621  0.630027   \n",
       "82412 -0.175305  0.519388 -0.052233 -0.118592 -0.150916  0.275564  0.715271   \n",
       "5114  -0.092799  0.680919  0.216972 -0.148479 -0.508639  0.453912  1.507544   \n",
       "17137 -0.007254  0.543765 -0.094489 -0.185421 -0.325657  0.197056  0.800147   \n",
       "98280  0.178256  0.564976 -0.289335 -0.127779  0.071391  0.310104  0.547856   \n",
       "\n",
       "       EMBED_21  EMBED_22  EMBED_23  EMBED_24  EMBED_25  EMBED_26  EMBED_27  \\\n",
       "71722 -0.357316  0.291080  0.354838 -0.001738  0.175875 -0.071201 -0.069151   \n",
       "82412 -0.157016  0.192690  0.159444  0.211376  0.152046  0.033970 -0.332181   \n",
       "5114   0.010893  0.603871  0.415800  0.221951  0.363705 -0.236890 -0.648948   \n",
       "17137 -0.115078  0.351718  0.373786  0.168323  0.222242 -0.088686 -0.163844   \n",
       "98280  0.163881  0.295819  0.224137 -0.278696  0.081467 -0.283003 -0.263821   \n",
       "\n",
       "       EMBED_28  EMBED_29  EMBED_30  EMBED_31  EMBED_32  EMBED_33  EMBED_34  \\\n",
       "71722  0.142517 -0.114328  0.233246  0.152829  0.109970 -0.527347  0.316943   \n",
       "82412  0.054970  0.003761  0.155690  0.253198  0.376423 -0.429835  0.224837   \n",
       "5114   0.115380  0.434274  0.357161  0.041595  0.505122 -0.471156  0.900400   \n",
       "17137  0.183858  0.086042  0.087384  0.203285  0.382363 -0.635049  0.420093   \n",
       "98280  0.184960  0.266168  0.057932 -0.132066  0.259967 -0.287199  0.383855   \n",
       "\n",
       "       EMBED_35  EMBED_36  EMBED_37  EMBED_38  EMBED_39  EMBED_40  EMBED_41  \\\n",
       "71722  0.050168 -0.174813  0.018598 -0.063050 -0.246874 -0.049857 -0.081755   \n",
       "82412 -0.067574 -0.071792 -0.275832  0.122419 -0.173609 -0.457992 -0.161361   \n",
       "5114  -0.452694 -0.166682 -0.275302 -0.109170  0.014999 -0.648206 -0.278187   \n",
       "17137 -0.106260 -0.208328 -0.035605  0.197733 -0.205399 -0.301760 -0.278275   \n",
       "98280 -0.101945 -0.196355  0.088330  0.207922 -0.158872 -0.079107 -0.076817   \n",
       "\n",
       "       EMBED_42  EMBED_43  EMBED_44  EMBED_45  EMBED_46  EMBED_47  EMBED_48  \\\n",
       "71722  0.008914 -0.299153 -0.541124  0.022232  0.144702  0.223001 -0.427878   \n",
       "82412 -0.088098 -0.041622 -0.774563  0.168907  0.237651  0.141014 -0.585243   \n",
       "5114  -0.153674 -0.036007 -0.217740 -0.103751  0.473294 -0.017797 -0.383454   \n",
       "17137 -0.378255 -0.034065 -0.539690 -0.064060  0.140130 -0.144546 -0.702111   \n",
       "98280 -0.413078  0.060489 -0.457690  0.310161 -0.064966  0.035300 -0.506857   \n",
       "\n",
       "       EMBED_49  EMBED_50  EMBED_51  EMBED_52  EMBED_53  EMBED_54  EMBED_55  \\\n",
       "71722  0.175729 -0.437079 -0.455686  0.011744 -0.188377 -0.098222  0.186742   \n",
       "82412 -0.089969 -0.441000 -0.412005  0.238198 -0.322015 -0.049432  0.081350   \n",
       "5114   0.001974 -0.505449  0.087838  0.134639 -0.086668  0.364900  0.125638   \n",
       "17137  0.107039 -0.250343 -0.245399  0.247315 -0.089495  0.025982  0.262099   \n",
       "98280 -0.017609 -0.121277 -0.549041 -0.130149 -0.372267 -0.348622  0.538688   \n",
       "\n",
       "       EMBED_56  EMBED_57  EMBED_58  EMBED_59  EMBED_60  EMBED_61  EMBED_62  \\\n",
       "71722  0.007582 -0.007565 -0.099802 -0.263568  0.166673  0.264614 -0.297953   \n",
       "82412 -0.264637 -0.287227 -0.276944 -0.359161  0.062582  0.092121 -0.136310   \n",
       "5114  -0.328315 -0.014870 -0.419891 -0.190609 -0.222180  0.154361  0.249562   \n",
       "17137 -0.102170 -0.187821 -0.167515 -0.331180 -0.017649  0.010149 -0.069578   \n",
       "98280 -0.004939 -0.150087 -0.305914 -0.177322  0.138986 -0.098812  0.206911   \n",
       "\n",
       "       EMBED_63  EMBED_64  EMBED_65  EMBED_66  EMBED_67  EMBED_68  EMBED_69  \\\n",
       "71722  0.126591 -0.294990  0.034109 -0.291699  0.186357 -0.169276  0.559821   \n",
       "82412  0.139558 -0.323911 -0.021034 -0.185649  0.255453 -0.063558  0.737593   \n",
       "5114   0.358595 -0.260617  0.187446 -0.156096  0.264043 -0.625066  0.065797   \n",
       "17137  0.314224 -0.188463 -0.055250 -0.186396  0.497756 -0.265870  0.462227   \n",
       "98280  0.385512 -0.516083 -0.156244 -0.157140  0.357336 -0.147980  0.238990   \n",
       "\n",
       "       EMBED_70  EMBED_71  EMBED_72  EMBED_73  EMBED_74  ...  EMBED_693  \\\n",
       "71722 -0.136239  0.065150  0.435436  0.272192  0.122566  ...  -0.074757   \n",
       "82412 -0.062536 -0.007713  0.547863  0.469338  0.166559  ...  -0.111920   \n",
       "5114   0.065141 -0.072569  0.474023  0.146674  0.593366  ...   0.167698   \n",
       "17137  0.008109 -0.003724  0.591356  0.405798  0.459157  ...  -0.184273   \n",
       "98280 -0.167907  0.039524  0.636579  0.384875  0.457279  ...  -0.138446   \n",
       "\n",
       "       EMBED_694  EMBED_695  EMBED_696  EMBED_697  EMBED_698  EMBED_699  \\\n",
       "71722  -0.100806  -0.128544   0.023274  -0.284944  -0.659821   0.027020   \n",
       "82412  -0.257764  -0.061592   0.127675  -0.116015  -0.591214  -0.203812   \n",
       "5114   -0.485078   0.149991  -0.262486  -0.443924  -0.202830  -0.599570   \n",
       "17137  -0.105833   0.005536   0.116708  -0.101085  -0.498121   0.005574   \n",
       "98280  -0.120921  -0.183172   0.215548   0.123629  -0.479293   0.029013   \n",
       "\n",
       "       EMBED_700  EMBED_701  EMBED_702  EMBED_703  EMBED_704  EMBED_705  \\\n",
       "71722   0.525081  -0.761540   0.068248  -0.071535  -0.207143  -0.391398   \n",
       "82412   0.636429  -0.498779  -0.087474  -0.318900  -0.270757  -0.250299   \n",
       "5114    0.992166  -0.565094  -0.437035   0.200376   0.090037  -0.136571   \n",
       "17137   0.775227  -0.391825   0.032070  -0.277728  -0.439826  -0.271632   \n",
       "98280   0.311060  -0.414726   0.218956  -0.479729  -0.387204  -0.573244   \n",
       "\n",
       "       EMBED_706  EMBED_707  EMBED_708  EMBED_709  EMBED_710  EMBED_711  \\\n",
       "71722   0.365011  -0.041412  -0.347234   0.137552  -0.297956   0.055337   \n",
       "82412   0.688977  -0.062132  -0.351459  -0.502622  -0.393250  -0.065938   \n",
       "5114    1.011193  -0.069717  -0.165624  -0.296528  -0.951483   0.021219   \n",
       "17137   0.670824  -0.046978  -0.446914  -0.233220  -0.424722   0.022346   \n",
       "98280   0.474536  -0.132899  -0.246493   0.091711  -0.274285  -0.162845   \n",
       "\n",
       "       EMBED_712  EMBED_713  EMBED_714  EMBED_715  EMBED_716  EMBED_717  \\\n",
       "71722   0.616171  -0.387249   0.127927   0.400709  -0.261731  -0.044241   \n",
       "82412   0.630741  -0.412833  -0.060410   0.451292  -0.372684  -0.172211   \n",
       "5114    0.649837  -0.118610  -0.220282   0.632864  -0.539973  -0.127614   \n",
       "17137   0.634328  -0.181058   0.138583   0.563455  -0.497472  -0.095692   \n",
       "98280   0.533131  -0.743020   0.276702   0.706932  -0.541503   0.082545   \n",
       "\n",
       "       EMBED_718  EMBED_719  EMBED_720  EMBED_721  EMBED_722  EMBED_723  \\\n",
       "71722  -0.150619   0.193708   0.128232   0.226463  -0.086620  -0.095651   \n",
       "82412  -0.214286   0.407914   0.042372   0.319015  -0.167069   0.006642   \n",
       "5114   -0.196914  -0.287379   0.224510  -0.198480  -0.553790   0.165419   \n",
       "17137  -0.403126   0.205138   0.078813   0.323774  -0.172926  -0.181271   \n",
       "98280  -0.194933   0.461641   0.093110   0.463067   0.137677  -0.174352   \n",
       "\n",
       "       EMBED_724  EMBED_725  EMBED_726  EMBED_727  EMBED_728  EMBED_729  \\\n",
       "71722  -0.471604  -0.369074  -0.236608  -0.208466  -0.676335  -0.396741   \n",
       "82412  -0.352651  -0.094960  -0.064348  -0.154181  -0.592284  -0.539266   \n",
       "5114   -0.146234  -0.197197  -0.251285  -0.137213  -0.186979  -0.730872   \n",
       "17137  -0.288377  -0.299346  -0.488301  -0.061735  -0.507592  -0.529043   \n",
       "98280  -0.150621  -0.439288  -0.506649  -0.567832  -0.723359  -0.870768   \n",
       "\n",
       "       EMBED_730  EMBED_731  EMBED_732  EMBED_733  EMBED_734  EMBED_735  \\\n",
       "71722   0.048930  -0.177669  -0.209846  -0.297194  -0.242546   0.043388   \n",
       "82412   0.355212   0.099570  -0.217730  -0.284940  -0.216200   0.059145   \n",
       "5114    0.166875   0.242221   0.085243  -0.731259   0.172389  -0.324964   \n",
       "17137   0.189368  -0.014910  -0.092923  -0.169689  -0.115287   0.007615   \n",
       "98280  -0.040746   0.053004  -0.174838   0.270660  -0.299528  -0.017289   \n",
       "\n",
       "       EMBED_736  EMBED_737  EMBED_738  EMBED_739  EMBED_740  EMBED_741  \\\n",
       "71722   0.051675   0.110699  -0.286888   0.320427  -0.144408   0.121379   \n",
       "82412   0.097196  -0.305455  -0.279240   0.402879   0.214084   0.278291   \n",
       "5114   -0.262773  -0.214583  -0.027991   0.414697   0.245987   0.057638   \n",
       "17137   0.133118  -0.055452  -0.275068   0.560709   0.234336  -0.057133   \n",
       "98280   0.525816  -0.245545   0.095457   0.530235  -0.098032  -0.017754   \n",
       "\n",
       "       EMBED_742  EMBED_743  EMBED_744  EMBED_745  EMBED_746  EMBED_747  \\\n",
       "71722   0.426598  -0.109743  -0.043318   0.287685   0.194216   0.350567   \n",
       "82412   0.127696   0.136121  -0.066565   0.282321   0.198715   0.477652   \n",
       "5114    0.437427   0.161853   0.030441   0.244832   0.240373   0.684028   \n",
       "17137   0.114402  -0.006842  -0.142725   0.400997   0.196825   0.314641   \n",
       "98280   0.219984   0.148890  -0.016947   0.300868   0.194767   0.113305   \n",
       "\n",
       "       EMBED_748  EMBED_749  EMBED_750  EMBED_751  EMBED_752  EMBED_753  \\\n",
       "71722   0.007508  -0.344494   0.273991   0.323298   0.683664   0.023797   \n",
       "82412   0.217438  -0.242581   0.337069   0.310483   0.532939  -0.021243   \n",
       "5114    0.251944   0.229340   0.176840   0.306029  -0.189102   0.068314   \n",
       "17137   0.263499  -0.360935   0.245796   0.327293   0.540557  -0.048752   \n",
       "98280   0.148186  -0.247089   0.237505   0.325652   0.214886  -0.149150   \n",
       "\n",
       "       EMBED_754  EMBED_755  EMBED_756  EMBED_757  EMBED_758  EMBED_759  \\\n",
       "71722  -0.576798   0.000249  -0.021221  -0.588558  -0.190467   0.273815   \n",
       "82412  -0.505779  -0.126655  -0.135804  -0.814749  -0.441541   0.225606   \n",
       "5114   -0.267470  -0.357423  -0.671371  -0.518839  -0.506295   0.155106   \n",
       "17137  -0.483916  -0.162530   0.071318  -0.695274  -0.291078   0.106476   \n",
       "98280  -0.274708   0.031862  -0.309526  -0.674247  -0.339018   0.068102   \n",
       "\n",
       "       EMBED_760  EMBED_761  EMBED_762  EMBED_763  EMBED_764  EMBED_765  \\\n",
       "71722  -0.097742  -0.292752  -0.022243  -0.217447  -0.979703   0.294535   \n",
       "82412  -0.030038  -0.071902   0.058272  -0.225572  -0.901695   0.082120   \n",
       "5114   -0.004346   0.238540   0.260278  -0.108356  -1.002016   0.154130   \n",
       "17137   0.135865   0.052395   0.025286  -0.130536  -1.048716   0.207678   \n",
       "98280   0.615809  -0.321040   0.283644  -0.118222  -1.054533  -0.016321   \n",
       "\n",
       "       EMBED_766  EMBED_767  \n",
       "71722  -0.416701   0.165263  \n",
       "82412  -0.124130  -0.104794  \n",
       "5114    0.441950  -0.105442  \n",
       "17137  -0.290715  -0.235323  \n",
       "98280  -0.325125   0.209312  \n",
       "\n",
       "[5 rows x 768 columns]"
      ]
     },
     "execution_count": 120,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "embed_df_cord.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training with embeds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 240,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_cord, target = build_df_target(diag_final, meds_final, labs_final, feats_to_merge_left=[embed_df_cord])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [],
   "source": [
    "cl = RandomForestClassifier(random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 242,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_cord = train_and_predict(df_cord, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 243,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.874\n",
      "\n",
      "LAB_51006_mean    0.007615\n",
      "LAB_51006_max     0.006489\n",
      "LAB_50912_mean    0.005896\n",
      "LAB_50912_max     0.005618\n",
      "LAB_51277_max     0.005458\n",
      "LAB_51277_mean    0.004777\n",
      "LAB_50910_min     0.003822\n",
      "LAB_51256_max     0.003744\n",
      "LAB_50931_max     0.003520\n",
      "LAB_51277_min     0.003462\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred_cord, cl, show_top_feats=10, col_names=df_cord.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top 500 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 251,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 6887), transformed shape: (31165, 500)\n",
      "CPU times: user 1min 13s, sys: 2.59 s, total: 1min 16s\n",
      "Wall time: 1min 17s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df_cord, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 252,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.871\n",
      "\n",
      "LAB_51006_max      0.020130\n",
      "LAB_51006_mean     0.017016\n",
      "LAB_51277_max      0.010024\n",
      "LAB_51277_mean     0.009921\n",
      "ICD9_4280_count    0.008542\n",
      "LAB_51249_min      0.008096\n",
      "LAB_50912_max      0.007751\n",
      "LAB_50912_mean     0.007244\n",
      "LAB_51277_min      0.006646\n",
      "LAB_51256_max      0.006426\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top features based on selection strategy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 253,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 6887), transformed shape: (31165, 1255)\n",
      "CPU times: user 1min 46s, sys: 1.89 s, total: 1min 48s\n",
      "Wall time: 1min 49s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df_cord, target, cl, threshold=None, max_feats=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 254,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.875\n",
      "\n",
      "LAB_51006_mean     0.012413\n",
      "LAB_51006_max      0.010281\n",
      "LAB_50912_max      0.008683\n",
      "LAB_51277_mean     0.008572\n",
      "LAB_51277_max      0.007507\n",
      "ICD9_4280_count    0.007372\n",
      "LAB_50912_mean     0.007016\n",
      "LAB_51249_min      0.005519\n",
      "EMBED_100          0.005501\n",
      "EMBED_341          0.004595\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add transformer embeddings for last discharge note of patients"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "DISCHARGE_NOTE_EMBEDS_PATH = \"../data/embeddings_discharge_history/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = AutoTokenizer.from_pretrained(\"deepset/covid_bert_base\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "config = AutoConfig.from_pretrained('deepset/covid_bert_base', output_hidden_states=True, output_attentions=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "model = AutoModelForMaskedLM.from_pretrained(\"deepset/covid_bert_base\", config=config)\n",
    "model_structure = model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "calculate_and_store_embeds(DISCHARGE_NOTE_EMBEDS_PATH,\n",
    "                           tokenizer, model,\n",
    "                           discharge_notes,\n",
    "                           text_col='HISTORY')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "embed_df = get_embeds_df(DISCHARGE_NOTE_EMBEDS_PATH)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "embed_df.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training with embeds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df, target = build_df_target(diag_final, meds_final, labs_final, feats_to_merge_left=[embed_df])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "cl = RandomForestClassifier(random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = train_and_predict(df, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, show_top_feats=10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Using bluebert transformer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 125,
   "metadata": {},
   "outputs": [],
   "source": [
    "BLUEBERT_LAST_NOTE_EMBEDS = \"../data/embeds_bluebert/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 126,
   "metadata": {},
   "outputs": [],
   "source": [
    "tokenizer = AutoTokenizer.from_pretrained(\"bionlp/bluebert_pubmed_uncased_L-24_H-1024_A-16\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "metadata": {},
   "outputs": [],
   "source": [
    "config = AutoConfig.from_pretrained(\"bionlp/bluebert_pubmed_uncased_L-24_H-1024_A-16\", output_hidden_states=True, output_attentions=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Some weights of the model checkpoint at bionlp/bluebert_pubmed_uncased_L-24_H-1024_A-16 were not used when initializing BertForMaskedLM: ['cls.seq_relationship.bias', 'cls.seq_relationship.weight']\n",
      "- This IS expected if you are initializing BertForMaskedLM from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
      "- This IS NOT expected if you are initializing BertForMaskedLM from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n"
     ]
    }
   ],
   "source": [
    "model = AutoModelForMaskedLM.from_pretrained(\"bionlp/bluebert_pubmed_uncased_L-24_H-1024_A-16\", config=config)\n",
    "model_structure = model.eval()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Asking to pad to max_length but no maximum length is provided and the model has no predefined maximum length. Default to no padding.\n",
      "Asking to truncate to max_length but no maximum length is provided and the model has no predefined maximum length. Default to no truncation.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 7.72 s, sys: 412 ms, total: 8.13 s\n",
      "Wall time: 2.11 s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "sentence = get_vector_for_text(last_note.TO_TOK.iloc[0], model, tokenizer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "cea780e9f8ec4ab29efd831a70bb8fee",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/38214 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "calculate_and_store_embeds(BLUEBERT_LAST_NOTE_EMBEDS,\n",
    "                           tokenizer, model,\n",
    "                           last_note,\n",
    "                           text_col='TO_TOK',\n",
    "                           max_length=512)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 131,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "application/vnd.jupyter.widget-view+json": {
       "model_id": "2c32a3dc8dc249f9a63193402943be65",
       "version_major": 2,
       "version_minor": 0
      },
      "text/plain": [
       "  0%|          | 0/38214 [00:00<?, ?it/s]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "embed_df_blue = get_embeds_df(BLUEBERT_LAST_NOTE_EMBEDS)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 132,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>EMBED_0</th>\n",
       "      <th>EMBED_1</th>\n",
       "      <th>EMBED_2</th>\n",
       "      <th>EMBED_3</th>\n",
       "      <th>EMBED_4</th>\n",
       "      <th>EMBED_5</th>\n",
       "      <th>EMBED_6</th>\n",
       "      <th>EMBED_7</th>\n",
       "      <th>EMBED_8</th>\n",
       "      <th>EMBED_9</th>\n",
       "      <th>EMBED_10</th>\n",
       "      <th>EMBED_11</th>\n",
       "      <th>EMBED_12</th>\n",
       "      <th>EMBED_13</th>\n",
       "      <th>EMBED_14</th>\n",
       "      <th>EMBED_15</th>\n",
       "      <th>EMBED_16</th>\n",
       "      <th>EMBED_17</th>\n",
       "      <th>EMBED_18</th>\n",
       "      <th>EMBED_19</th>\n",
       "      <th>EMBED_20</th>\n",
       "      <th>EMBED_21</th>\n",
       "      <th>EMBED_22</th>\n",
       "      <th>EMBED_23</th>\n",
       "      <th>EMBED_24</th>\n",
       "      <th>EMBED_25</th>\n",
       "      <th>EMBED_26</th>\n",
       "      <th>EMBED_27</th>\n",
       "      <th>EMBED_28</th>\n",
       "      <th>EMBED_29</th>\n",
       "      <th>EMBED_30</th>\n",
       "      <th>EMBED_31</th>\n",
       "      <th>EMBED_32</th>\n",
       "      <th>EMBED_33</th>\n",
       "      <th>EMBED_34</th>\n",
       "      <th>EMBED_35</th>\n",
       "      <th>EMBED_36</th>\n",
       "      <th>EMBED_37</th>\n",
       "      <th>EMBED_38</th>\n",
       "      <th>EMBED_39</th>\n",
       "      <th>EMBED_40</th>\n",
       "      <th>EMBED_41</th>\n",
       "      <th>EMBED_42</th>\n",
       "      <th>EMBED_43</th>\n",
       "      <th>EMBED_44</th>\n",
       "      <th>EMBED_45</th>\n",
       "      <th>EMBED_46</th>\n",
       "      <th>EMBED_47</th>\n",
       "      <th>EMBED_48</th>\n",
       "      <th>EMBED_49</th>\n",
       "      <th>EMBED_50</th>\n",
       "      <th>EMBED_51</th>\n",
       "      <th>EMBED_52</th>\n",
       "      <th>EMBED_53</th>\n",
       "      <th>EMBED_54</th>\n",
       "      <th>EMBED_55</th>\n",
       "      <th>EMBED_56</th>\n",
       "      <th>EMBED_57</th>\n",
       "      <th>EMBED_58</th>\n",
       "      <th>EMBED_59</th>\n",
       "      <th>EMBED_60</th>\n",
       "      <th>EMBED_61</th>\n",
       "      <th>EMBED_62</th>\n",
       "      <th>EMBED_63</th>\n",
       "      <th>EMBED_64</th>\n",
       "      <th>EMBED_65</th>\n",
       "      <th>EMBED_66</th>\n",
       "      <th>EMBED_67</th>\n",
       "      <th>EMBED_68</th>\n",
       "      <th>EMBED_69</th>\n",
       "      <th>EMBED_70</th>\n",
       "      <th>EMBED_71</th>\n",
       "      <th>EMBED_72</th>\n",
       "      <th>EMBED_73</th>\n",
       "      <th>EMBED_74</th>\n",
       "      <th>...</th>\n",
       "      <th>EMBED_949</th>\n",
       "      <th>EMBED_950</th>\n",
       "      <th>EMBED_951</th>\n",
       "      <th>EMBED_952</th>\n",
       "      <th>EMBED_953</th>\n",
       "      <th>EMBED_954</th>\n",
       "      <th>EMBED_955</th>\n",
       "      <th>EMBED_956</th>\n",
       "      <th>EMBED_957</th>\n",
       "      <th>EMBED_958</th>\n",
       "      <th>EMBED_959</th>\n",
       "      <th>EMBED_960</th>\n",
       "      <th>EMBED_961</th>\n",
       "      <th>EMBED_962</th>\n",
       "      <th>EMBED_963</th>\n",
       "      <th>EMBED_964</th>\n",
       "      <th>EMBED_965</th>\n",
       "      <th>EMBED_966</th>\n",
       "      <th>EMBED_967</th>\n",
       "      <th>EMBED_968</th>\n",
       "      <th>EMBED_969</th>\n",
       "      <th>EMBED_970</th>\n",
       "      <th>EMBED_971</th>\n",
       "      <th>EMBED_972</th>\n",
       "      <th>EMBED_973</th>\n",
       "      <th>EMBED_974</th>\n",
       "      <th>EMBED_975</th>\n",
       "      <th>EMBED_976</th>\n",
       "      <th>EMBED_977</th>\n",
       "      <th>EMBED_978</th>\n",
       "      <th>EMBED_979</th>\n",
       "      <th>EMBED_980</th>\n",
       "      <th>EMBED_981</th>\n",
       "      <th>EMBED_982</th>\n",
       "      <th>EMBED_983</th>\n",
       "      <th>EMBED_984</th>\n",
       "      <th>EMBED_985</th>\n",
       "      <th>EMBED_986</th>\n",
       "      <th>EMBED_987</th>\n",
       "      <th>EMBED_988</th>\n",
       "      <th>EMBED_989</th>\n",
       "      <th>EMBED_990</th>\n",
       "      <th>EMBED_991</th>\n",
       "      <th>EMBED_992</th>\n",
       "      <th>EMBED_993</th>\n",
       "      <th>EMBED_994</th>\n",
       "      <th>EMBED_995</th>\n",
       "      <th>EMBED_996</th>\n",
       "      <th>EMBED_997</th>\n",
       "      <th>EMBED_998</th>\n",
       "      <th>EMBED_999</th>\n",
       "      <th>EMBED_1000</th>\n",
       "      <th>EMBED_1001</th>\n",
       "      <th>EMBED_1002</th>\n",
       "      <th>EMBED_1003</th>\n",
       "      <th>EMBED_1004</th>\n",
       "      <th>EMBED_1005</th>\n",
       "      <th>EMBED_1006</th>\n",
       "      <th>EMBED_1007</th>\n",
       "      <th>EMBED_1008</th>\n",
       "      <th>EMBED_1009</th>\n",
       "      <th>EMBED_1010</th>\n",
       "      <th>EMBED_1011</th>\n",
       "      <th>EMBED_1012</th>\n",
       "      <th>EMBED_1013</th>\n",
       "      <th>EMBED_1014</th>\n",
       "      <th>EMBED_1015</th>\n",
       "      <th>EMBED_1016</th>\n",
       "      <th>EMBED_1017</th>\n",
       "      <th>EMBED_1018</th>\n",
       "      <th>EMBED_1019</th>\n",
       "      <th>EMBED_1020</th>\n",
       "      <th>EMBED_1021</th>\n",
       "      <th>EMBED_1022</th>\n",
       "      <th>EMBED_1023</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>71722</th>\n",
       "      <td>0.126159</td>\n",
       "      <td>0.436510</td>\n",
       "      <td>0.242167</td>\n",
       "      <td>0.075487</td>\n",
       "      <td>-0.066005</td>\n",
       "      <td>0.160982</td>\n",
       "      <td>-0.556105</td>\n",
       "      <td>-0.116901</td>\n",
       "      <td>-0.044531</td>\n",
       "      <td>0.035461</td>\n",
       "      <td>0.214184</td>\n",
       "      <td>-0.093669</td>\n",
       "      <td>-0.162777</td>\n",
       "      <td>0.060817</td>\n",
       "      <td>-0.389239</td>\n",
       "      <td>0.424134</td>\n",
       "      <td>0.061676</td>\n",
       "      <td>-0.559476</td>\n",
       "      <td>-0.536042</td>\n",
       "      <td>-0.114947</td>\n",
       "      <td>-0.299029</td>\n",
       "      <td>0.227966</td>\n",
       "      <td>-0.100967</td>\n",
       "      <td>0.503610</td>\n",
       "      <td>-0.265138</td>\n",
       "      <td>0.130300</td>\n",
       "      <td>0.079612</td>\n",
       "      <td>0.032571</td>\n",
       "      <td>0.060839</td>\n",
       "      <td>0.404185</td>\n",
       "      <td>-0.464865</td>\n",
       "      <td>0.062208</td>\n",
       "      <td>0.762042</td>\n",
       "      <td>0.112792</td>\n",
       "      <td>-0.118231</td>\n",
       "      <td>0.356843</td>\n",
       "      <td>0.253962</td>\n",
       "      <td>-0.536314</td>\n",
       "      <td>-0.852881</td>\n",
       "      <td>-0.314916</td>\n",
       "      <td>0.170798</td>\n",
       "      <td>0.377525</td>\n",
       "      <td>-0.221571</td>\n",
       "      <td>0.036960</td>\n",
       "      <td>-0.424111</td>\n",
       "      <td>-0.096845</td>\n",
       "      <td>-0.319887</td>\n",
       "      <td>-0.225143</td>\n",
       "      <td>0.062301</td>\n",
       "      <td>-0.211184</td>\n",
       "      <td>0.151822</td>\n",
       "      <td>-0.172018</td>\n",
       "      <td>0.013553</td>\n",
       "      <td>-0.102084</td>\n",
       "      <td>-0.846182</td>\n",
       "      <td>-0.003040</td>\n",
       "      <td>-0.296922</td>\n",
       "      <td>-0.326186</td>\n",
       "      <td>-0.219615</td>\n",
       "      <td>0.252852</td>\n",
       "      <td>0.241332</td>\n",
       "      <td>0.610302</td>\n",
       "      <td>0.120713</td>\n",
       "      <td>-0.245511</td>\n",
       "      <td>0.081097</td>\n",
       "      <td>-0.132424</td>\n",
       "      <td>0.309821</td>\n",
       "      <td>-0.034169</td>\n",
       "      <td>0.159709</td>\n",
       "      <td>-0.044685</td>\n",
       "      <td>-0.173835</td>\n",
       "      <td>0.019423</td>\n",
       "      <td>0.041678</td>\n",
       "      <td>0.095126</td>\n",
       "      <td>0.110174</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.011611</td>\n",
       "      <td>-0.327034</td>\n",
       "      <td>-0.406547</td>\n",
       "      <td>0.070501</td>\n",
       "      <td>0.385579</td>\n",
       "      <td>0.730299</td>\n",
       "      <td>0.382424</td>\n",
       "      <td>-0.456475</td>\n",
       "      <td>-0.039866</td>\n",
       "      <td>-0.152783</td>\n",
       "      <td>0.322712</td>\n",
       "      <td>-0.294549</td>\n",
       "      <td>0.122282</td>\n",
       "      <td>-0.471311</td>\n",
       "      <td>-0.155821</td>\n",
       "      <td>-0.296040</td>\n",
       "      <td>-0.042809</td>\n",
       "      <td>0.001762</td>\n",
       "      <td>0.217919</td>\n",
       "      <td>-0.202362</td>\n",
       "      <td>0.219308</td>\n",
       "      <td>-0.241887</td>\n",
       "      <td>0.283836</td>\n",
       "      <td>0.076513</td>\n",
       "      <td>0.185603</td>\n",
       "      <td>-0.154774</td>\n",
       "      <td>0.289012</td>\n",
       "      <td>0.320288</td>\n",
       "      <td>0.098252</td>\n",
       "      <td>-0.271678</td>\n",
       "      <td>0.020878</td>\n",
       "      <td>-0.239889</td>\n",
       "      <td>0.288305</td>\n",
       "      <td>-0.439668</td>\n",
       "      <td>-0.521628</td>\n",
       "      <td>-0.134196</td>\n",
       "      <td>0.082636</td>\n",
       "      <td>-0.154364</td>\n",
       "      <td>0.142685</td>\n",
       "      <td>-0.166480</td>\n",
       "      <td>0.039682</td>\n",
       "      <td>-0.016251</td>\n",
       "      <td>-0.346280</td>\n",
       "      <td>0.547977</td>\n",
       "      <td>-0.238038</td>\n",
       "      <td>0.486565</td>\n",
       "      <td>0.517330</td>\n",
       "      <td>-0.068962</td>\n",
       "      <td>-0.133956</td>\n",
       "      <td>-0.096405</td>\n",
       "      <td>0.315044</td>\n",
       "      <td>0.370287</td>\n",
       "      <td>-0.038405</td>\n",
       "      <td>-0.101288</td>\n",
       "      <td>0.495852</td>\n",
       "      <td>0.179552</td>\n",
       "      <td>0.216966</td>\n",
       "      <td>0.062044</td>\n",
       "      <td>0.202901</td>\n",
       "      <td>-0.105785</td>\n",
       "      <td>0.020583</td>\n",
       "      <td>0.565039</td>\n",
       "      <td>-0.219493</td>\n",
       "      <td>-0.198222</td>\n",
       "      <td>0.012717</td>\n",
       "      <td>-0.622802</td>\n",
       "      <td>-0.408033</td>\n",
       "      <td>-0.361794</td>\n",
       "      <td>-0.327248</td>\n",
       "      <td>0.594166</td>\n",
       "      <td>0.363368</td>\n",
       "      <td>0.285408</td>\n",
       "      <td>0.072183</td>\n",
       "      <td>-0.126077</td>\n",
       "      <td>-0.738355</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>82412</th>\n",
       "      <td>-0.033801</td>\n",
       "      <td>0.239327</td>\n",
       "      <td>0.009834</td>\n",
       "      <td>-0.095445</td>\n",
       "      <td>0.006841</td>\n",
       "      <td>-0.098751</td>\n",
       "      <td>-0.177360</td>\n",
       "      <td>0.085988</td>\n",
       "      <td>0.154912</td>\n",
       "      <td>0.218953</td>\n",
       "      <td>0.286586</td>\n",
       "      <td>0.040244</td>\n",
       "      <td>-0.018348</td>\n",
       "      <td>0.312531</td>\n",
       "      <td>-0.393073</td>\n",
       "      <td>-0.023888</td>\n",
       "      <td>0.086215</td>\n",
       "      <td>-0.580380</td>\n",
       "      <td>-0.324074</td>\n",
       "      <td>-0.162556</td>\n",
       "      <td>-0.497023</td>\n",
       "      <td>0.174009</td>\n",
       "      <td>-0.376378</td>\n",
       "      <td>0.187240</td>\n",
       "      <td>-0.209004</td>\n",
       "      <td>0.117070</td>\n",
       "      <td>0.031478</td>\n",
       "      <td>0.180658</td>\n",
       "      <td>-0.012397</td>\n",
       "      <td>0.207855</td>\n",
       "      <td>-0.577716</td>\n",
       "      <td>-0.091582</td>\n",
       "      <td>0.534677</td>\n",
       "      <td>0.192129</td>\n",
       "      <td>0.041386</td>\n",
       "      <td>0.178551</td>\n",
       "      <td>0.474371</td>\n",
       "      <td>-0.348173</td>\n",
       "      <td>-0.638711</td>\n",
       "      <td>-0.175002</td>\n",
       "      <td>0.064390</td>\n",
       "      <td>0.337212</td>\n",
       "      <td>0.072763</td>\n",
       "      <td>-0.152635</td>\n",
       "      <td>-0.378980</td>\n",
       "      <td>-0.111861</td>\n",
       "      <td>-0.361720</td>\n",
       "      <td>0.041414</td>\n",
       "      <td>0.123014</td>\n",
       "      <td>-0.090510</td>\n",
       "      <td>-0.080031</td>\n",
       "      <td>-0.021958</td>\n",
       "      <td>0.232678</td>\n",
       "      <td>-0.317472</td>\n",
       "      <td>-0.653503</td>\n",
       "      <td>-0.029335</td>\n",
       "      <td>-0.357550</td>\n",
       "      <td>-0.166688</td>\n",
       "      <td>-0.070455</td>\n",
       "      <td>-0.137656</td>\n",
       "      <td>-0.050704</td>\n",
       "      <td>0.462457</td>\n",
       "      <td>0.063943</td>\n",
       "      <td>-0.291309</td>\n",
       "      <td>-0.123869</td>\n",
       "      <td>-0.187115</td>\n",
       "      <td>0.013233</td>\n",
       "      <td>-0.078966</td>\n",
       "      <td>0.336568</td>\n",
       "      <td>-0.066077</td>\n",
       "      <td>-0.484040</td>\n",
       "      <td>-0.096028</td>\n",
       "      <td>-0.145613</td>\n",
       "      <td>0.054264</td>\n",
       "      <td>-0.046829</td>\n",
       "      <td>...</td>\n",
       "      <td>0.037423</td>\n",
       "      <td>-0.408514</td>\n",
       "      <td>-0.472092</td>\n",
       "      <td>0.167567</td>\n",
       "      <td>0.134359</td>\n",
       "      <td>0.600675</td>\n",
       "      <td>0.267510</td>\n",
       "      <td>-0.496247</td>\n",
       "      <td>-0.064671</td>\n",
       "      <td>0.135613</td>\n",
       "      <td>0.261699</td>\n",
       "      <td>-0.292142</td>\n",
       "      <td>-0.016971</td>\n",
       "      <td>-0.419180</td>\n",
       "      <td>-0.211325</td>\n",
       "      <td>-0.314253</td>\n",
       "      <td>-0.184657</td>\n",
       "      <td>0.457590</td>\n",
       "      <td>0.378102</td>\n",
       "      <td>-0.410953</td>\n",
       "      <td>0.082616</td>\n",
       "      <td>0.150130</td>\n",
       "      <td>0.702130</td>\n",
       "      <td>0.234030</td>\n",
       "      <td>0.132476</td>\n",
       "      <td>-0.247872</td>\n",
       "      <td>0.155987</td>\n",
       "      <td>0.332855</td>\n",
       "      <td>-0.181230</td>\n",
       "      <td>-0.129482</td>\n",
       "      <td>-0.037377</td>\n",
       "      <td>-0.182101</td>\n",
       "      <td>0.338332</td>\n",
       "      <td>-0.530658</td>\n",
       "      <td>-0.677319</td>\n",
       "      <td>-0.233210</td>\n",
       "      <td>0.159586</td>\n",
       "      <td>-0.075922</td>\n",
       "      <td>0.189328</td>\n",
       "      <td>-0.204555</td>\n",
       "      <td>0.042659</td>\n",
       "      <td>0.164566</td>\n",
       "      <td>-0.550907</td>\n",
       "      <td>0.561801</td>\n",
       "      <td>-0.304060</td>\n",
       "      <td>0.311331</td>\n",
       "      <td>0.342475</td>\n",
       "      <td>0.035392</td>\n",
       "      <td>0.090733</td>\n",
       "      <td>-0.071335</td>\n",
       "      <td>0.118521</td>\n",
       "      <td>0.341236</td>\n",
       "      <td>-0.192608</td>\n",
       "      <td>0.122883</td>\n",
       "      <td>0.394619</td>\n",
       "      <td>0.718032</td>\n",
       "      <td>-0.132527</td>\n",
       "      <td>0.304222</td>\n",
       "      <td>0.406020</td>\n",
       "      <td>0.029856</td>\n",
       "      <td>0.059273</td>\n",
       "      <td>0.359587</td>\n",
       "      <td>-0.550240</td>\n",
       "      <td>-0.323758</td>\n",
       "      <td>0.136708</td>\n",
       "      <td>-0.003329</td>\n",
       "      <td>-0.219950</td>\n",
       "      <td>-0.600653</td>\n",
       "      <td>-0.283652</td>\n",
       "      <td>0.239964</td>\n",
       "      <td>0.034882</td>\n",
       "      <td>0.303956</td>\n",
       "      <td>0.246657</td>\n",
       "      <td>-0.204607</td>\n",
       "      <td>-0.555502</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5114</th>\n",
       "      <td>-0.191577</td>\n",
       "      <td>-0.036818</td>\n",
       "      <td>0.114636</td>\n",
       "      <td>-0.222577</td>\n",
       "      <td>-0.249242</td>\n",
       "      <td>-0.263088</td>\n",
       "      <td>-0.142193</td>\n",
       "      <td>-0.116250</td>\n",
       "      <td>0.201194</td>\n",
       "      <td>0.419393</td>\n",
       "      <td>-0.001027</td>\n",
       "      <td>0.101421</td>\n",
       "      <td>-0.184931</td>\n",
       "      <td>0.011129</td>\n",
       "      <td>-0.117897</td>\n",
       "      <td>0.020725</td>\n",
       "      <td>0.347723</td>\n",
       "      <td>-0.402812</td>\n",
       "      <td>0.212848</td>\n",
       "      <td>-0.239313</td>\n",
       "      <td>-0.138923</td>\n",
       "      <td>0.031540</td>\n",
       "      <td>-0.195069</td>\n",
       "      <td>-0.406853</td>\n",
       "      <td>-0.378589</td>\n",
       "      <td>0.405224</td>\n",
       "      <td>0.000920</td>\n",
       "      <td>-0.008358</td>\n",
       "      <td>-0.014336</td>\n",
       "      <td>0.260850</td>\n",
       "      <td>-0.361500</td>\n",
       "      <td>0.048678</td>\n",
       "      <td>0.494480</td>\n",
       "      <td>0.015765</td>\n",
       "      <td>0.109617</td>\n",
       "      <td>0.368569</td>\n",
       "      <td>0.292078</td>\n",
       "      <td>0.030244</td>\n",
       "      <td>-0.392454</td>\n",
       "      <td>0.078273</td>\n",
       "      <td>0.119657</td>\n",
       "      <td>0.274546</td>\n",
       "      <td>0.040932</td>\n",
       "      <td>-0.205556</td>\n",
       "      <td>-0.303096</td>\n",
       "      <td>0.251245</td>\n",
       "      <td>-0.668638</td>\n",
       "      <td>-0.026762</td>\n",
       "      <td>0.266305</td>\n",
       "      <td>0.105079</td>\n",
       "      <td>-0.171233</td>\n",
       "      <td>0.180575</td>\n",
       "      <td>0.118146</td>\n",
       "      <td>0.076032</td>\n",
       "      <td>-0.417952</td>\n",
       "      <td>-0.254436</td>\n",
       "      <td>-0.269481</td>\n",
       "      <td>0.014096</td>\n",
       "      <td>0.010788</td>\n",
       "      <td>-0.030079</td>\n",
       "      <td>-0.053799</td>\n",
       "      <td>0.327781</td>\n",
       "      <td>0.214395</td>\n",
       "      <td>-0.193339</td>\n",
       "      <td>-0.159697</td>\n",
       "      <td>-0.180567</td>\n",
       "      <td>-0.065915</td>\n",
       "      <td>-0.085503</td>\n",
       "      <td>0.446732</td>\n",
       "      <td>-0.138953</td>\n",
       "      <td>-0.291464</td>\n",
       "      <td>0.267119</td>\n",
       "      <td>0.073202</td>\n",
       "      <td>-0.166091</td>\n",
       "      <td>0.248310</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.218498</td>\n",
       "      <td>-0.421519</td>\n",
       "      <td>-0.129988</td>\n",
       "      <td>0.079731</td>\n",
       "      <td>0.174985</td>\n",
       "      <td>0.178628</td>\n",
       "      <td>0.054949</td>\n",
       "      <td>-0.160682</td>\n",
       "      <td>0.131846</td>\n",
       "      <td>0.205291</td>\n",
       "      <td>0.176385</td>\n",
       "      <td>-0.233475</td>\n",
       "      <td>0.032087</td>\n",
       "      <td>0.041352</td>\n",
       "      <td>0.260905</td>\n",
       "      <td>-0.039111</td>\n",
       "      <td>-0.412093</td>\n",
       "      <td>0.201381</td>\n",
       "      <td>0.135255</td>\n",
       "      <td>0.048177</td>\n",
       "      <td>0.174186</td>\n",
       "      <td>-0.076255</td>\n",
       "      <td>0.110815</td>\n",
       "      <td>0.097563</td>\n",
       "      <td>0.092943</td>\n",
       "      <td>-0.187329</td>\n",
       "      <td>0.049509</td>\n",
       "      <td>0.085749</td>\n",
       "      <td>-0.163091</td>\n",
       "      <td>0.027573</td>\n",
       "      <td>-0.078574</td>\n",
       "      <td>-0.527003</td>\n",
       "      <td>0.023994</td>\n",
       "      <td>-0.499332</td>\n",
       "      <td>-0.257219</td>\n",
       "      <td>0.048213</td>\n",
       "      <td>0.103565</td>\n",
       "      <td>0.088230</td>\n",
       "      <td>-0.015312</td>\n",
       "      <td>0.180714</td>\n",
       "      <td>0.056706</td>\n",
       "      <td>0.130726</td>\n",
       "      <td>-0.349609</td>\n",
       "      <td>0.381002</td>\n",
       "      <td>-0.159568</td>\n",
       "      <td>0.369587</td>\n",
       "      <td>-0.139158</td>\n",
       "      <td>-0.099504</td>\n",
       "      <td>0.033194</td>\n",
       "      <td>-0.224749</td>\n",
       "      <td>0.132486</td>\n",
       "      <td>0.083885</td>\n",
       "      <td>0.331759</td>\n",
       "      <td>-0.016488</td>\n",
       "      <td>0.095156</td>\n",
       "      <td>0.324151</td>\n",
       "      <td>0.125316</td>\n",
       "      <td>0.329509</td>\n",
       "      <td>0.244068</td>\n",
       "      <td>0.008921</td>\n",
       "      <td>-0.016191</td>\n",
       "      <td>-0.047972</td>\n",
       "      <td>-0.548679</td>\n",
       "      <td>-0.114532</td>\n",
       "      <td>-0.147883</td>\n",
       "      <td>-0.141304</td>\n",
       "      <td>-0.039461</td>\n",
       "      <td>-0.076981</td>\n",
       "      <td>-0.168604</td>\n",
       "      <td>0.223234</td>\n",
       "      <td>0.394777</td>\n",
       "      <td>-0.188741</td>\n",
       "      <td>-0.133674</td>\n",
       "      <td>-0.006437</td>\n",
       "      <td>-0.201993</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17137</th>\n",
       "      <td>-0.211029</td>\n",
       "      <td>0.237688</td>\n",
       "      <td>-0.054573</td>\n",
       "      <td>-0.038889</td>\n",
       "      <td>0.287835</td>\n",
       "      <td>0.128505</td>\n",
       "      <td>-0.405968</td>\n",
       "      <td>0.053832</td>\n",
       "      <td>0.380447</td>\n",
       "      <td>0.680215</td>\n",
       "      <td>0.034469</td>\n",
       "      <td>0.115404</td>\n",
       "      <td>-0.061639</td>\n",
       "      <td>-0.309535</td>\n",
       "      <td>-0.357594</td>\n",
       "      <td>-0.106217</td>\n",
       "      <td>0.086009</td>\n",
       "      <td>-0.611335</td>\n",
       "      <td>-0.110591</td>\n",
       "      <td>-0.009403</td>\n",
       "      <td>-0.324664</td>\n",
       "      <td>-0.230553</td>\n",
       "      <td>-0.173653</td>\n",
       "      <td>-0.139129</td>\n",
       "      <td>-0.683387</td>\n",
       "      <td>0.296482</td>\n",
       "      <td>-0.027812</td>\n",
       "      <td>0.079788</td>\n",
       "      <td>-0.033823</td>\n",
       "      <td>0.399768</td>\n",
       "      <td>-0.526472</td>\n",
       "      <td>0.019377</td>\n",
       "      <td>0.534149</td>\n",
       "      <td>0.441749</td>\n",
       "      <td>-0.446372</td>\n",
       "      <td>0.688240</td>\n",
       "      <td>0.414382</td>\n",
       "      <td>0.192735</td>\n",
       "      <td>-0.628935</td>\n",
       "      <td>-0.209340</td>\n",
       "      <td>0.023467</td>\n",
       "      <td>0.478772</td>\n",
       "      <td>0.062231</td>\n",
       "      <td>-0.029898</td>\n",
       "      <td>-0.157493</td>\n",
       "      <td>-0.227254</td>\n",
       "      <td>-0.051262</td>\n",
       "      <td>0.126902</td>\n",
       "      <td>-0.229894</td>\n",
       "      <td>0.100410</td>\n",
       "      <td>0.348955</td>\n",
       "      <td>-0.017569</td>\n",
       "      <td>0.120301</td>\n",
       "      <td>-0.540708</td>\n",
       "      <td>-0.615325</td>\n",
       "      <td>0.264917</td>\n",
       "      <td>-0.342038</td>\n",
       "      <td>0.003423</td>\n",
       "      <td>0.093901</td>\n",
       "      <td>0.213414</td>\n",
       "      <td>0.411046</td>\n",
       "      <td>0.823031</td>\n",
       "      <td>0.238755</td>\n",
       "      <td>-0.050477</td>\n",
       "      <td>-0.102609</td>\n",
       "      <td>-0.066173</td>\n",
       "      <td>0.267656</td>\n",
       "      <td>0.098945</td>\n",
       "      <td>0.345777</td>\n",
       "      <td>-0.433650</td>\n",
       "      <td>-0.144763</td>\n",
       "      <td>-0.025979</td>\n",
       "      <td>0.117753</td>\n",
       "      <td>-0.253587</td>\n",
       "      <td>-0.023717</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.128127</td>\n",
       "      <td>-0.249281</td>\n",
       "      <td>-0.516097</td>\n",
       "      <td>0.328534</td>\n",
       "      <td>0.019446</td>\n",
       "      <td>0.460966</td>\n",
       "      <td>0.135424</td>\n",
       "      <td>-0.379272</td>\n",
       "      <td>0.064263</td>\n",
       "      <td>0.472445</td>\n",
       "      <td>0.570769</td>\n",
       "      <td>-0.203336</td>\n",
       "      <td>0.132061</td>\n",
       "      <td>-0.524748</td>\n",
       "      <td>-0.028132</td>\n",
       "      <td>0.235888</td>\n",
       "      <td>-0.189899</td>\n",
       "      <td>0.380986</td>\n",
       "      <td>0.124684</td>\n",
       "      <td>-0.227342</td>\n",
       "      <td>0.240454</td>\n",
       "      <td>-0.002464</td>\n",
       "      <td>0.311112</td>\n",
       "      <td>0.007978</td>\n",
       "      <td>-0.011209</td>\n",
       "      <td>-0.445169</td>\n",
       "      <td>0.353669</td>\n",
       "      <td>0.282584</td>\n",
       "      <td>0.185076</td>\n",
       "      <td>-0.294659</td>\n",
       "      <td>-0.287108</td>\n",
       "      <td>-0.246007</td>\n",
       "      <td>0.099570</td>\n",
       "      <td>-0.385953</td>\n",
       "      <td>-0.465863</td>\n",
       "      <td>-0.205500</td>\n",
       "      <td>0.326308</td>\n",
       "      <td>-0.041224</td>\n",
       "      <td>-0.029949</td>\n",
       "      <td>-0.182842</td>\n",
       "      <td>0.285433</td>\n",
       "      <td>0.041454</td>\n",
       "      <td>-0.650968</td>\n",
       "      <td>0.509722</td>\n",
       "      <td>-0.178976</td>\n",
       "      <td>0.178410</td>\n",
       "      <td>0.440970</td>\n",
       "      <td>-0.200375</td>\n",
       "      <td>-0.114586</td>\n",
       "      <td>-0.163598</td>\n",
       "      <td>0.368468</td>\n",
       "      <td>0.290797</td>\n",
       "      <td>0.083558</td>\n",
       "      <td>-0.213727</td>\n",
       "      <td>0.342349</td>\n",
       "      <td>0.585226</td>\n",
       "      <td>0.045634</td>\n",
       "      <td>0.494608</td>\n",
       "      <td>0.422412</td>\n",
       "      <td>-0.036288</td>\n",
       "      <td>-0.119138</td>\n",
       "      <td>0.105387</td>\n",
       "      <td>-0.607848</td>\n",
       "      <td>0.009303</td>\n",
       "      <td>-0.317806</td>\n",
       "      <td>-0.289958</td>\n",
       "      <td>-0.145019</td>\n",
       "      <td>-0.257415</td>\n",
       "      <td>-0.640967</td>\n",
       "      <td>0.531011</td>\n",
       "      <td>0.176170</td>\n",
       "      <td>0.244654</td>\n",
       "      <td>-0.114302</td>\n",
       "      <td>-0.332156</td>\n",
       "      <td>-0.733184</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98280</th>\n",
       "      <td>-0.073435</td>\n",
       "      <td>0.144499</td>\n",
       "      <td>0.126824</td>\n",
       "      <td>-0.512236</td>\n",
       "      <td>0.181070</td>\n",
       "      <td>-0.358486</td>\n",
       "      <td>-0.436533</td>\n",
       "      <td>0.324561</td>\n",
       "      <td>0.290541</td>\n",
       "      <td>-0.097479</td>\n",
       "      <td>0.099323</td>\n",
       "      <td>0.032851</td>\n",
       "      <td>0.189333</td>\n",
       "      <td>0.237339</td>\n",
       "      <td>-0.848396</td>\n",
       "      <td>0.104682</td>\n",
       "      <td>-0.149326</td>\n",
       "      <td>-0.569303</td>\n",
       "      <td>-0.225219</td>\n",
       "      <td>-0.184061</td>\n",
       "      <td>-0.438047</td>\n",
       "      <td>0.140051</td>\n",
       "      <td>0.066204</td>\n",
       "      <td>0.299454</td>\n",
       "      <td>-0.341741</td>\n",
       "      <td>0.110075</td>\n",
       "      <td>-0.097543</td>\n",
       "      <td>0.325448</td>\n",
       "      <td>-0.280757</td>\n",
       "      <td>-0.079213</td>\n",
       "      <td>-0.990128</td>\n",
       "      <td>-0.039309</td>\n",
       "      <td>0.469903</td>\n",
       "      <td>-0.022016</td>\n",
       "      <td>0.417348</td>\n",
       "      <td>0.644910</td>\n",
       "      <td>0.339572</td>\n",
       "      <td>-0.815207</td>\n",
       "      <td>-0.809533</td>\n",
       "      <td>-0.140182</td>\n",
       "      <td>-0.219720</td>\n",
       "      <td>0.429991</td>\n",
       "      <td>-0.105468</td>\n",
       "      <td>-0.090842</td>\n",
       "      <td>-0.304628</td>\n",
       "      <td>-0.432353</td>\n",
       "      <td>-0.392002</td>\n",
       "      <td>0.404235</td>\n",
       "      <td>0.140739</td>\n",
       "      <td>0.125102</td>\n",
       "      <td>0.194765</td>\n",
       "      <td>0.164811</td>\n",
       "      <td>-0.379786</td>\n",
       "      <td>-0.102278</td>\n",
       "      <td>-0.700691</td>\n",
       "      <td>-0.146675</td>\n",
       "      <td>0.121066</td>\n",
       "      <td>-0.273050</td>\n",
       "      <td>-0.099012</td>\n",
       "      <td>-0.381656</td>\n",
       "      <td>-0.004443</td>\n",
       "      <td>0.572652</td>\n",
       "      <td>-0.147982</td>\n",
       "      <td>-0.459262</td>\n",
       "      <td>-0.203447</td>\n",
       "      <td>0.169294</td>\n",
       "      <td>-0.008218</td>\n",
       "      <td>0.081470</td>\n",
       "      <td>0.478125</td>\n",
       "      <td>-0.187945</td>\n",
       "      <td>-0.600027</td>\n",
       "      <td>-0.147246</td>\n",
       "      <td>-0.107341</td>\n",
       "      <td>0.182777</td>\n",
       "      <td>-0.134620</td>\n",
       "      <td>...</td>\n",
       "      <td>0.134751</td>\n",
       "      <td>-0.604807</td>\n",
       "      <td>-0.213321</td>\n",
       "      <td>-0.059350</td>\n",
       "      <td>0.433602</td>\n",
       "      <td>0.474274</td>\n",
       "      <td>0.458703</td>\n",
       "      <td>-0.521393</td>\n",
       "      <td>-0.142062</td>\n",
       "      <td>0.099244</td>\n",
       "      <td>0.921301</td>\n",
       "      <td>0.043233</td>\n",
       "      <td>0.318978</td>\n",
       "      <td>-0.281421</td>\n",
       "      <td>-0.179311</td>\n",
       "      <td>-0.888624</td>\n",
       "      <td>-0.169406</td>\n",
       "      <td>0.104738</td>\n",
       "      <td>0.524390</td>\n",
       "      <td>0.060558</td>\n",
       "      <td>-0.362997</td>\n",
       "      <td>0.112607</td>\n",
       "      <td>0.203717</td>\n",
       "      <td>0.476431</td>\n",
       "      <td>-0.084053</td>\n",
       "      <td>-0.026530</td>\n",
       "      <td>0.355641</td>\n",
       "      <td>-0.058962</td>\n",
       "      <td>-0.248482</td>\n",
       "      <td>-0.065345</td>\n",
       "      <td>0.001641</td>\n",
       "      <td>-0.102789</td>\n",
       "      <td>0.368618</td>\n",
       "      <td>-0.072713</td>\n",
       "      <td>-0.953267</td>\n",
       "      <td>-0.339503</td>\n",
       "      <td>0.254983</td>\n",
       "      <td>-0.260511</td>\n",
       "      <td>-0.171908</td>\n",
       "      <td>0.311264</td>\n",
       "      <td>-0.021756</td>\n",
       "      <td>-0.306776</td>\n",
       "      <td>-0.744168</td>\n",
       "      <td>0.466079</td>\n",
       "      <td>-0.356645</td>\n",
       "      <td>0.635172</td>\n",
       "      <td>0.514110</td>\n",
       "      <td>-0.175560</td>\n",
       "      <td>-0.368511</td>\n",
       "      <td>-0.032713</td>\n",
       "      <td>-0.048887</td>\n",
       "      <td>0.425029</td>\n",
       "      <td>-0.049640</td>\n",
       "      <td>0.236184</td>\n",
       "      <td>0.357695</td>\n",
       "      <td>0.264326</td>\n",
       "      <td>0.040066</td>\n",
       "      <td>-0.148775</td>\n",
       "      <td>0.261721</td>\n",
       "      <td>0.099788</td>\n",
       "      <td>-0.217441</td>\n",
       "      <td>0.795276</td>\n",
       "      <td>-0.936652</td>\n",
       "      <td>-0.384704</td>\n",
       "      <td>0.276310</td>\n",
       "      <td>-0.361896</td>\n",
       "      <td>-0.055852</td>\n",
       "      <td>-0.413039</td>\n",
       "      <td>-0.698286</td>\n",
       "      <td>0.808524</td>\n",
       "      <td>0.487466</td>\n",
       "      <td>0.118813</td>\n",
       "      <td>0.277577</td>\n",
       "      <td>-0.110008</td>\n",
       "      <td>-0.352216</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 1024 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        EMBED_0   EMBED_1   EMBED_2   EMBED_3   EMBED_4   EMBED_5   EMBED_6  \\\n",
       "71722  0.126159  0.436510  0.242167  0.075487 -0.066005  0.160982 -0.556105   \n",
       "82412 -0.033801  0.239327  0.009834 -0.095445  0.006841 -0.098751 -0.177360   \n",
       "5114  -0.191577 -0.036818  0.114636 -0.222577 -0.249242 -0.263088 -0.142193   \n",
       "17137 -0.211029  0.237688 -0.054573 -0.038889  0.287835  0.128505 -0.405968   \n",
       "98280 -0.073435  0.144499  0.126824 -0.512236  0.181070 -0.358486 -0.436533   \n",
       "\n",
       "        EMBED_7   EMBED_8   EMBED_9  EMBED_10  EMBED_11  EMBED_12  EMBED_13  \\\n",
       "71722 -0.116901 -0.044531  0.035461  0.214184 -0.093669 -0.162777  0.060817   \n",
       "82412  0.085988  0.154912  0.218953  0.286586  0.040244 -0.018348  0.312531   \n",
       "5114  -0.116250  0.201194  0.419393 -0.001027  0.101421 -0.184931  0.011129   \n",
       "17137  0.053832  0.380447  0.680215  0.034469  0.115404 -0.061639 -0.309535   \n",
       "98280  0.324561  0.290541 -0.097479  0.099323  0.032851  0.189333  0.237339   \n",
       "\n",
       "       EMBED_14  EMBED_15  EMBED_16  EMBED_17  EMBED_18  EMBED_19  EMBED_20  \\\n",
       "71722 -0.389239  0.424134  0.061676 -0.559476 -0.536042 -0.114947 -0.299029   \n",
       "82412 -0.393073 -0.023888  0.086215 -0.580380 -0.324074 -0.162556 -0.497023   \n",
       "5114  -0.117897  0.020725  0.347723 -0.402812  0.212848 -0.239313 -0.138923   \n",
       "17137 -0.357594 -0.106217  0.086009 -0.611335 -0.110591 -0.009403 -0.324664   \n",
       "98280 -0.848396  0.104682 -0.149326 -0.569303 -0.225219 -0.184061 -0.438047   \n",
       "\n",
       "       EMBED_21  EMBED_22  EMBED_23  EMBED_24  EMBED_25  EMBED_26  EMBED_27  \\\n",
       "71722  0.227966 -0.100967  0.503610 -0.265138  0.130300  0.079612  0.032571   \n",
       "82412  0.174009 -0.376378  0.187240 -0.209004  0.117070  0.031478  0.180658   \n",
       "5114   0.031540 -0.195069 -0.406853 -0.378589  0.405224  0.000920 -0.008358   \n",
       "17137 -0.230553 -0.173653 -0.139129 -0.683387  0.296482 -0.027812  0.079788   \n",
       "98280  0.140051  0.066204  0.299454 -0.341741  0.110075 -0.097543  0.325448   \n",
       "\n",
       "       EMBED_28  EMBED_29  EMBED_30  EMBED_31  EMBED_32  EMBED_33  EMBED_34  \\\n",
       "71722  0.060839  0.404185 -0.464865  0.062208  0.762042  0.112792 -0.118231   \n",
       "82412 -0.012397  0.207855 -0.577716 -0.091582  0.534677  0.192129  0.041386   \n",
       "5114  -0.014336  0.260850 -0.361500  0.048678  0.494480  0.015765  0.109617   \n",
       "17137 -0.033823  0.399768 -0.526472  0.019377  0.534149  0.441749 -0.446372   \n",
       "98280 -0.280757 -0.079213 -0.990128 -0.039309  0.469903 -0.022016  0.417348   \n",
       "\n",
       "       EMBED_35  EMBED_36  EMBED_37  EMBED_38  EMBED_39  EMBED_40  EMBED_41  \\\n",
       "71722  0.356843  0.253962 -0.536314 -0.852881 -0.314916  0.170798  0.377525   \n",
       "82412  0.178551  0.474371 -0.348173 -0.638711 -0.175002  0.064390  0.337212   \n",
       "5114   0.368569  0.292078  0.030244 -0.392454  0.078273  0.119657  0.274546   \n",
       "17137  0.688240  0.414382  0.192735 -0.628935 -0.209340  0.023467  0.478772   \n",
       "98280  0.644910  0.339572 -0.815207 -0.809533 -0.140182 -0.219720  0.429991   \n",
       "\n",
       "       EMBED_42  EMBED_43  EMBED_44  EMBED_45  EMBED_46  EMBED_47  EMBED_48  \\\n",
       "71722 -0.221571  0.036960 -0.424111 -0.096845 -0.319887 -0.225143  0.062301   \n",
       "82412  0.072763 -0.152635 -0.378980 -0.111861 -0.361720  0.041414  0.123014   \n",
       "5114   0.040932 -0.205556 -0.303096  0.251245 -0.668638 -0.026762  0.266305   \n",
       "17137  0.062231 -0.029898 -0.157493 -0.227254 -0.051262  0.126902 -0.229894   \n",
       "98280 -0.105468 -0.090842 -0.304628 -0.432353 -0.392002  0.404235  0.140739   \n",
       "\n",
       "       EMBED_49  EMBED_50  EMBED_51  EMBED_52  EMBED_53  EMBED_54  EMBED_55  \\\n",
       "71722 -0.211184  0.151822 -0.172018  0.013553 -0.102084 -0.846182 -0.003040   \n",
       "82412 -0.090510 -0.080031 -0.021958  0.232678 -0.317472 -0.653503 -0.029335   \n",
       "5114   0.105079 -0.171233  0.180575  0.118146  0.076032 -0.417952 -0.254436   \n",
       "17137  0.100410  0.348955 -0.017569  0.120301 -0.540708 -0.615325  0.264917   \n",
       "98280  0.125102  0.194765  0.164811 -0.379786 -0.102278 -0.700691 -0.146675   \n",
       "\n",
       "       EMBED_56  EMBED_57  EMBED_58  EMBED_59  EMBED_60  EMBED_61  EMBED_62  \\\n",
       "71722 -0.296922 -0.326186 -0.219615  0.252852  0.241332  0.610302  0.120713   \n",
       "82412 -0.357550 -0.166688 -0.070455 -0.137656 -0.050704  0.462457  0.063943   \n",
       "5114  -0.269481  0.014096  0.010788 -0.030079 -0.053799  0.327781  0.214395   \n",
       "17137 -0.342038  0.003423  0.093901  0.213414  0.411046  0.823031  0.238755   \n",
       "98280  0.121066 -0.273050 -0.099012 -0.381656 -0.004443  0.572652 -0.147982   \n",
       "\n",
       "       EMBED_63  EMBED_64  EMBED_65  EMBED_66  EMBED_67  EMBED_68  EMBED_69  \\\n",
       "71722 -0.245511  0.081097 -0.132424  0.309821 -0.034169  0.159709 -0.044685   \n",
       "82412 -0.291309 -0.123869 -0.187115  0.013233 -0.078966  0.336568 -0.066077   \n",
       "5114  -0.193339 -0.159697 -0.180567 -0.065915 -0.085503  0.446732 -0.138953   \n",
       "17137 -0.050477 -0.102609 -0.066173  0.267656  0.098945  0.345777 -0.433650   \n",
       "98280 -0.459262 -0.203447  0.169294 -0.008218  0.081470  0.478125 -0.187945   \n",
       "\n",
       "       EMBED_70  EMBED_71  EMBED_72  EMBED_73  EMBED_74  ...  EMBED_949  \\\n",
       "71722 -0.173835  0.019423  0.041678  0.095126  0.110174  ...  -0.011611   \n",
       "82412 -0.484040 -0.096028 -0.145613  0.054264 -0.046829  ...   0.037423   \n",
       "5114  -0.291464  0.267119  0.073202 -0.166091  0.248310  ...  -0.218498   \n",
       "17137 -0.144763 -0.025979  0.117753 -0.253587 -0.023717  ...  -0.128127   \n",
       "98280 -0.600027 -0.147246 -0.107341  0.182777 -0.134620  ...   0.134751   \n",
       "\n",
       "       EMBED_950  EMBED_951  EMBED_952  EMBED_953  EMBED_954  EMBED_955  \\\n",
       "71722  -0.327034  -0.406547   0.070501   0.385579   0.730299   0.382424   \n",
       "82412  -0.408514  -0.472092   0.167567   0.134359   0.600675   0.267510   \n",
       "5114   -0.421519  -0.129988   0.079731   0.174985   0.178628   0.054949   \n",
       "17137  -0.249281  -0.516097   0.328534   0.019446   0.460966   0.135424   \n",
       "98280  -0.604807  -0.213321  -0.059350   0.433602   0.474274   0.458703   \n",
       "\n",
       "       EMBED_956  EMBED_957  EMBED_958  EMBED_959  EMBED_960  EMBED_961  \\\n",
       "71722  -0.456475  -0.039866  -0.152783   0.322712  -0.294549   0.122282   \n",
       "82412  -0.496247  -0.064671   0.135613   0.261699  -0.292142  -0.016971   \n",
       "5114   -0.160682   0.131846   0.205291   0.176385  -0.233475   0.032087   \n",
       "17137  -0.379272   0.064263   0.472445   0.570769  -0.203336   0.132061   \n",
       "98280  -0.521393  -0.142062   0.099244   0.921301   0.043233   0.318978   \n",
       "\n",
       "       EMBED_962  EMBED_963  EMBED_964  EMBED_965  EMBED_966  EMBED_967  \\\n",
       "71722  -0.471311  -0.155821  -0.296040  -0.042809   0.001762   0.217919   \n",
       "82412  -0.419180  -0.211325  -0.314253  -0.184657   0.457590   0.378102   \n",
       "5114    0.041352   0.260905  -0.039111  -0.412093   0.201381   0.135255   \n",
       "17137  -0.524748  -0.028132   0.235888  -0.189899   0.380986   0.124684   \n",
       "98280  -0.281421  -0.179311  -0.888624  -0.169406   0.104738   0.524390   \n",
       "\n",
       "       EMBED_968  EMBED_969  EMBED_970  EMBED_971  EMBED_972  EMBED_973  \\\n",
       "71722  -0.202362   0.219308  -0.241887   0.283836   0.076513   0.185603   \n",
       "82412  -0.410953   0.082616   0.150130   0.702130   0.234030   0.132476   \n",
       "5114    0.048177   0.174186  -0.076255   0.110815   0.097563   0.092943   \n",
       "17137  -0.227342   0.240454  -0.002464   0.311112   0.007978  -0.011209   \n",
       "98280   0.060558  -0.362997   0.112607   0.203717   0.476431  -0.084053   \n",
       "\n",
       "       EMBED_974  EMBED_975  EMBED_976  EMBED_977  EMBED_978  EMBED_979  \\\n",
       "71722  -0.154774   0.289012   0.320288   0.098252  -0.271678   0.020878   \n",
       "82412  -0.247872   0.155987   0.332855  -0.181230  -0.129482  -0.037377   \n",
       "5114   -0.187329   0.049509   0.085749  -0.163091   0.027573  -0.078574   \n",
       "17137  -0.445169   0.353669   0.282584   0.185076  -0.294659  -0.287108   \n",
       "98280  -0.026530   0.355641  -0.058962  -0.248482  -0.065345   0.001641   \n",
       "\n",
       "       EMBED_980  EMBED_981  EMBED_982  EMBED_983  EMBED_984  EMBED_985  \\\n",
       "71722  -0.239889   0.288305  -0.439668  -0.521628  -0.134196   0.082636   \n",
       "82412  -0.182101   0.338332  -0.530658  -0.677319  -0.233210   0.159586   \n",
       "5114   -0.527003   0.023994  -0.499332  -0.257219   0.048213   0.103565   \n",
       "17137  -0.246007   0.099570  -0.385953  -0.465863  -0.205500   0.326308   \n",
       "98280  -0.102789   0.368618  -0.072713  -0.953267  -0.339503   0.254983   \n",
       "\n",
       "       EMBED_986  EMBED_987  EMBED_988  EMBED_989  EMBED_990  EMBED_991  \\\n",
       "71722  -0.154364   0.142685  -0.166480   0.039682  -0.016251  -0.346280   \n",
       "82412  -0.075922   0.189328  -0.204555   0.042659   0.164566  -0.550907   \n",
       "5114    0.088230  -0.015312   0.180714   0.056706   0.130726  -0.349609   \n",
       "17137  -0.041224  -0.029949  -0.182842   0.285433   0.041454  -0.650968   \n",
       "98280  -0.260511  -0.171908   0.311264  -0.021756  -0.306776  -0.744168   \n",
       "\n",
       "       EMBED_992  EMBED_993  EMBED_994  EMBED_995  EMBED_996  EMBED_997  \\\n",
       "71722   0.547977  -0.238038   0.486565   0.517330  -0.068962  -0.133956   \n",
       "82412   0.561801  -0.304060   0.311331   0.342475   0.035392   0.090733   \n",
       "5114    0.381002  -0.159568   0.369587  -0.139158  -0.099504   0.033194   \n",
       "17137   0.509722  -0.178976   0.178410   0.440970  -0.200375  -0.114586   \n",
       "98280   0.466079  -0.356645   0.635172   0.514110  -0.175560  -0.368511   \n",
       "\n",
       "       EMBED_998  EMBED_999  EMBED_1000  EMBED_1001  EMBED_1002  EMBED_1003  \\\n",
       "71722  -0.096405   0.315044    0.370287   -0.038405   -0.101288    0.495852   \n",
       "82412  -0.071335   0.118521    0.341236   -0.192608    0.122883    0.394619   \n",
       "5114   -0.224749   0.132486    0.083885    0.331759   -0.016488    0.095156   \n",
       "17137  -0.163598   0.368468    0.290797    0.083558   -0.213727    0.342349   \n",
       "98280  -0.032713  -0.048887    0.425029   -0.049640    0.236184    0.357695   \n",
       "\n",
       "       EMBED_1004  EMBED_1005  EMBED_1006  EMBED_1007  EMBED_1008  EMBED_1009  \\\n",
       "71722    0.179552    0.216966    0.062044    0.202901   -0.105785    0.020583   \n",
       "82412    0.718032   -0.132527    0.304222    0.406020    0.029856    0.059273   \n",
       "5114     0.324151    0.125316    0.329509    0.244068    0.008921   -0.016191   \n",
       "17137    0.585226    0.045634    0.494608    0.422412   -0.036288   -0.119138   \n",
       "98280    0.264326    0.040066   -0.148775    0.261721    0.099788   -0.217441   \n",
       "\n",
       "       EMBED_1010  EMBED_1011  EMBED_1012  EMBED_1013  EMBED_1014  EMBED_1015  \\\n",
       "71722    0.565039   -0.219493   -0.198222    0.012717   -0.622802   -0.408033   \n",
       "82412    0.359587   -0.550240   -0.323758    0.136708   -0.003329   -0.219950   \n",
       "5114    -0.047972   -0.548679   -0.114532   -0.147883   -0.141304   -0.039461   \n",
       "17137    0.105387   -0.607848    0.009303   -0.317806   -0.289958   -0.145019   \n",
       "98280    0.795276   -0.936652   -0.384704    0.276310   -0.361896   -0.055852   \n",
       "\n",
       "       EMBED_1016  EMBED_1017  EMBED_1018  EMBED_1019  EMBED_1020  EMBED_1021  \\\n",
       "71722   -0.361794   -0.327248    0.594166    0.363368    0.285408    0.072183   \n",
       "82412   -0.600653   -0.283652    0.239964    0.034882    0.303956    0.246657   \n",
       "5114    -0.076981   -0.168604    0.223234    0.394777   -0.188741   -0.133674   \n",
       "17137   -0.257415   -0.640967    0.531011    0.176170    0.244654   -0.114302   \n",
       "98280   -0.413039   -0.698286    0.808524    0.487466    0.118813    0.277577   \n",
       "\n",
       "       EMBED_1022  EMBED_1023  \n",
       "71722   -0.126077   -0.738355  \n",
       "82412   -0.204607   -0.555502  \n",
       "5114    -0.006437   -0.201993  \n",
       "17137   -0.332156   -0.733184  \n",
       "98280   -0.110008   -0.352216  \n",
       "\n",
       "[5 rows x 1024 columns]"
      ]
     },
     "execution_count": 132,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "embed_df_blue.head()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Training with embeds"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 255,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_bb, target = build_df_target(diag_final, meds_final, labs_final, feats_to_merge_left=[embed_df_blue])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 256,
   "metadata": {},
   "outputs": [],
   "source": [
    "cl = RandomForestClassifier(random_state=RANDOM_SEED)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 257,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_bb = train_and_predict(df_bb, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 260,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.870\n",
      "\n",
      "LAB_51006_max      0.008446\n",
      "LAB_51277_max      0.005858\n",
      "LAB_51006_mean     0.005770\n",
      "LAB_51249_min      0.005756\n",
      "LAB_51277_mean     0.004966\n",
      "ICD9_4280_count    0.003725\n",
      "LAB_51222_mean     0.003614\n",
      "LAB_50931_mean     0.003509\n",
      "EMBED_327          0.003498\n",
      "LAB_51277_min      0.003322\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred_bb, cl, show_top_feats=10, col_names=df_bb.columns)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top 500 features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 261,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 7143), transformed shape: (31165, 500)\n",
      "CPU times: user 1min 10s, sys: 2.33 s, total: 1min 12s\n",
      "Wall time: 1min 13s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df_bb, target, cl)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 262,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.871\n",
      "\n",
      "LAB_51006_max      0.021531\n",
      "LAB_51006_mean     0.012972\n",
      "LAB_51277_max      0.012598\n",
      "LAB_51277_mean     0.011942\n",
      "LAB_51249_min      0.008894\n",
      "LAB_50912_max      0.008787\n",
      "LAB_51277_min      0.007462\n",
      "ICD9_4280_count    0.006659\n",
      "LAB_51256_max      0.005815\n",
      "LAB_50910_min      0.005722\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Prediction using top features based on selection strategy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 263,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Old shape: (31165, 7143), transformed shape: (31165, 1482)\n",
      "CPU times: user 1min 53s, sys: 3.08 s, total: 1min 56s\n",
      "Wall time: 1min 58s\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "pred, retrained_cl, new_feats = train_and_predict_with_select_feats(df_bb, target, cl, threshold=None, max_feats=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 264,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ROC-AUC: 0.870\n",
      "\n",
      "LAB_51006_mean     0.011058\n",
      "LAB_51277_max      0.008753\n",
      "LAB_51277_mean     0.008331\n",
      "LAB_51006_max      0.008295\n",
      "LAB_50912_mean     0.005241\n",
      "LAB_51277_min      0.004955\n",
      "LAB_51249_min      0.004903\n",
      "ICD9_4280_count    0.004777\n",
      "LAB_51218_max      0.004645\n",
      "EMBED_194          0.003879\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "show_results(target[target.index.isin(TEST_IDS)], pred, retrained_cl, show_top_feats=10, col_names=new_feats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Comparing respiratory diseases and other diseases"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [],
   "source": [
    "acute_diag_codes = [\n",
    "460, 461, 462, 463, 464, 465, 466\n",
    "]\n",
    "other_resp_tract_diag_codes = [\n",
    "470, 471, 472, 473, 474, 475, 476, 477, 478\n",
    "]\n",
    "pneumonia_and_influenza_diag_codes = [\n",
    "480, 481, 482, 483, 484, 485, 486, 487, 488\n",
    "]\n",
    "grp4 = [\n",
    "490, 491, 492, 493, 494, 495, 496\n",
    "]\n",
    "grp5 = [\n",
    "500, 501, 502, 503, 504, 505, 506, 507, 508\n",
    "]\n",
    "grp6 = [\n",
    "510, 511, 512, 513, 514, 515, 516, 517, 518, 519\n",
    "]\n",
    "\n",
    "RESPIRATORY_DIAG_CODES: List[int] = [*acute_diag_codes, *other_resp_tract_diag_codes, *pneumonia_and_influenza_diag_codes, *grp4, *grp5, *grp6]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [],
   "source": [
    "obs_wind_diags = diag_preprocessed[(diag_preprocessed.DATE < diag_preprocessed.SUBJECT_ID.map(last_date)) & (diag_preprocessed.DATE > diag_preprocessed.SUBJECT_ID.map(earliest_date))]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 148,
   "metadata": {},
   "outputs": [],
   "source": [
    "obs_wind_diags = obs_wind_diags[obs_wind_diags.FEATURE_NAME.notna()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 149,
   "metadata": {},
   "outputs": [],
   "source": [
    "def is_respiratory(diag: str) -> bool:\n",
    "    for code in RESPIRATORY_DIAG_CODES:\n",
    "        if diag.split('_')[-1].startswith(str(code)):\n",
    "            return True\n",
    "    return False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 150,
   "metadata": {},
   "outputs": [],
   "source": [
    "respiratory_diags = obs_wind_diags[obs_wind_diags.FEATURE_NAME.map(is_respiratory)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "metadata": {},
   "outputs": [],
   "source": [
    "respiratory_patients = set(respiratory_diags.SUBJECT_ID)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [],
   "source": [
    "comparison = pd.DataFrame([\n",
    "    target[target.index.isin(TEST_IDS)].to_list(), \n",
    "    list(pred_cord),\n",
    "    list(pred_bb),\n",
    "    target[target.index.isin(TEST_IDS)].index.isin(respiratory_patients)\n",
    "], index=['target', 'cord', 'bb', 'respiratory']).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 153,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "respiratory  target\n",
       "False        False     3990\n",
       "True         False     1953\n",
       "             True      1041\n",
       "False        True       762\n",
       "dtype: int64"
      ]
     },
     "execution_count": 153,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "comparison[['respiratory', 'target']].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 154,
   "metadata": {},
   "outputs": [],
   "source": [
    "def print_roc_aucs(df: pd.DataFrame):\n",
    "    print(f\"CORD ROC-AUC: {roc_auc_score(df.target.astype(bool), df.cord):.3f}, BlueBERT ROC-AUC: {roc_auc_score(df.target.astype(bool), df.bb):.3f}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [],
   "source": [
    "def find_roc_aucs_for_comparison(comparison: pd.DataFrame):\n",
    "    print(\"Respiratory:\")\n",
    "    print_roc_aucs(comparison[comparison.respiratory])\n",
    "    print(\"Non-respiratory:\")\n",
    "    print_roc_aucs(comparison[~comparison.respiratory])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.8256557929793962"
      ]
     },
     "execution_count": 156,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "roc_auc_score(comparison[comparison.respiratory].target.astype(bool), comparison[comparison.respiratory].cord)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Respiratory:\n",
      "CORD ROC-AUC: 0.826, BlueBERT ROC-AUC: 0.820\n",
      "Non-respiratory:\n",
      "CORD ROC-AUC: 0.892, BlueBERT ROC-AUC: 0.889\n"
     ]
    }
   ],
   "source": [
    "find_roc_aucs_for_comparison(comparison)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}