examples/eval-ddrsa.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "b80ff03c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-03-01T22:03:15.817497Z",
     "start_time": "2023-03-01T22:03:15.800683Z"
    },
    "execution": {
     "iopub.execute_input": "2023-04-24T07:24:05.790521Z",
     "iopub.status.busy": "2023-04-24T07:24:05.790293Z",
     "iopub.status.idle": "2023-04-24T07:24:05.797287Z",
     "shell.execute_reply": "2023-04-24T07:24:05.796630Z",
     "shell.execute_reply.started": "2023-04-24T07:24:05.790500Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "SEED = 1234\n",
    "_EPSILON = 1e-08"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "acc16d0a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-03-02T16:00:14.081365Z",
     "start_time": "2023-03-02T16:00:13.647673Z"
    },
    "execution": {
     "iopub.execute_input": "2023-04-24T07:24:06.673040Z",
     "iopub.status.busy": "2023-04-24T07:24:06.672807Z",
     "iopub.status.idle": "2023-04-24T07:24:09.694066Z",
     "shell.execute_reply": "2023-04-24T07:24:09.693270Z",
     "shell.execute_reply.started": "2023-04-24T07:24:06.673019Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import sys\n",
    "sys.path.append('../scr/')\n",
    "\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import torch.nn as nn\n",
    "import torch\n",
    "from torch.utils.data import DataLoader\n",
    "import torch.nn.functional as F\n",
    "import os\n",
    "from sklearn.model_selection import train_test_split\n",
    "from tqdm import tqdm\n",
    "from torch.utils.tensorboard import SummaryWriter\n",
    "import datetime\n",
    "\n",
    "from ddrsa import DDRSA\n",
    "import import_data as impt\n",
    "from utils_eval import c_index, brier_score\n",
    "from utils_helper import f_get_boosted_trainset\n",
    "from utils_helper import f_get_risk_predictions, _f_get_pred\n",
    "from utils_helper import save_checkpoint, load_checkpoint"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "b52a0d6c",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-03-02T01:52:48.981491Z",
     "start_time": "2023-03-02T01:52:48.954519Z"
    },
    "execution": {
     "iopub.execute_input": "2023-04-24T07:24:09.695548Z",
     "iopub.status.busy": "2023-04-24T07:24:09.695189Z",
     "iopub.status.idle": "2023-04-24T07:24:09.700683Z",
     "shell.execute_reply": "2023-04-24T07:24:09.700123Z",
     "shell.execute_reply.started": "2023-04-24T07:24:09.695523Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "## for reproducibility\n",
    "torch.manual_seed(SEED)\n",
    "torch.backends.cudnn.deterministic = True\n",
    "torch.backends.cudnn.benchmark = False\n",
    "np.random.seed(SEED)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1d41c04f-c0fc-403c-95a0-c43684e36c93",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Part I: Load the model"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c3cde6d",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Import Dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5f68d319",
   "metadata": {},
   "source": [
    "### load EEG for testing set 1234\n",
    "load data with high-resolution EEG data (better for visualization)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "a2a00f2a",
   "metadata": {},
   "outputs": [],
   "source": [
    "## if the combined data with high resolution EEG is already saved as a file, \n",
    "## load the test data from a saved file\n",
    "## if not, run the follow two cells\n",
    "import pickle\n",
    "with open('high_res_combined_data_test.pickle', 'rb') as handle:\n",
    "    te_all_eeg = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "da06e1da",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████| 185/185 [13:01<00:00,  4.22s/it]\n"
     ]
    }
   ],
   "source": [
    "## load the test data from raw\n",
    "## else, skip the following two cells\n",
    "import glob\n",
    "from tqdm import tqdm\n",
    "import math\n",
    "\n",
    "## testing patient list with random seed of 1234\n",
    "te_id_1234 = \\\n",
    "    ['PUH-2019-032', 'PUH-2011-125', 'PUH-2014-216', 'PUH-2014-187', 'PUH-2017-229', 'PUH-2018-046', 'PUH-2013-069', 'PUH-2018-164', 'PUH-2012-210', 'PUH-2015-122', 'PUH-2014-150', 'PUH-2011-163', 'PUH-2015-136', 'PUH-2016-092', 'PUH-2017-027', 'PUH-2013-140', 'PUH-2018-054', 'PUH-2014-174', 'PUH-2019-022', 'PUH-2019-037', 'PUH-2016-048', 'PUH-2014-130', 'PUH-2016-181', 'PUH-2012-158', 'PUH-2010-205', 'PUH-2015-095', 'PUH-2013-194', 'PUH-2011-166', 'PUH-2012-186', 'PUH-2014-211', 'PUH-2011-017', 'PUH-2016-175', 'PUH-2015-270', 'PUH-2013-063', 'PUH-2016-277', 'PUH-2017-296', 'PUH-2019-108', 'PUH-2016-313', 'PUH-2017-098', 'PUH-2011-089', 'PUH-2017-077', 'PUH-2011-071', 'PUH-2016-142', 'PUH-2017-057', 'PUH-2013-165', 'PUH-2016-304', 'PUH-2013-002', 'PUH-2011-154', 'PUH-2013-059', 'PUH-2018-302', 'PUH-2016-080', 'PUH-2018-042', 'PUH-2011-120', 'PUH-2018-167', 'PUH-2017-294', 'PUH-2016-246', 'PUH-2019-124', 'PUH-2016-100', 'PUH-2014-079', 'PUH-2015-293', 'PUH-2011-152', 'PUH-2016-027', 'PUH-2017-181', 'PUH-2012-043', 'PUH-2010-181', 'PUH-2016-257', 'PUH-2010-165', 'PUH-2011-030', 'PUH-2014-106', 'PUH-2012-095', 'PUH-2012-214', 'PUH-2016-242', 'PUH-2012-223', 'PUH-2011-111', 'PUH-2014-060', 'PUH-2013-032', 'PUH-2015-264', 'PUH-2015-207', 'PUH-2017-292', 'PUH-2015-051', 'PUH-2010-140', 'PUH-2014-088', 'PUH-2013-172', 'PUH-2016-324', 'PUH-2015-258', 'PUH-2014-253', 'PUH-2015-295', 'PUH-2017-044', 'PUH-2016-203', 'PUH-2015-019', 'PUH-2017-096', 'PUH-2016-052', 'PUH-2012-083', 'PUH-2018-078', 'PUH-2016-084', 'PUH-2012-032', 'PUH-2013-099', 'PUH-2016-329', 'PUH-2014-249', 'PUH-2010-207', 'PUH-2012-066', 'PUH-2013-061', 'PUH-2012-197', 'PUH-2019-091', 'PUH-2011-075', 'PUH-2017-138', 'PUH-2017-093', 'PUH-2013-166', 'PUH-2010-208', 'PUH-2010-169', 'PUH-2013-208', 'PUH-2015-124', 'PUH-2013-034', 'PUH-2019-059', 'PUH-2015-244', 'PUH-2014-251', 'PUH-2014-227', 'PUH-2013-183', 'PUH-2015-086', 'PUH-2012-188', 'PUH-2017-021', 'PUH-2015-044', 'PUH-2012-183', 'PUH-2015-080', 'PUH-2010-158', 'PUH-2017-011', 'PUH-2013-199', 'PUH-2015-030', 'PUH-2014-093', 'PUH-2010-180', 'PUH-2013-021', 'PUH-2014-241', 'PUH-2016-025', 'PUH-2014-248', 'PUH-2010-150', 'PUH-2010-177', 'PUH-2014-168', 'PUH-2014-184', 'PUH-2018-131', 'PUH-2018-322', 'PUH-2017-213', 'PUH-2015-009', 'PUH-2018-117', 'PUH-2015-282', 'PUH-2013-177', 'PUH-2013-142', 'PUH-2018-245', 'PUH-2016-297', 'PUH-2010-076', 'PUH-2011-139', 'PUH-2014-021', 'PUH-2018-126', 'PUH-2019-040', 'PUH-2013-111', 'PUH-2016-147', 'PUH-2013-019', 'PUH-2013-001', 'PUH-2017-278', 'PUH-2016-248', 'PUH-2014-236', 'PUH-2017-087', 'PUH-2014-083', 'PUH-2016-135', 'PUH-2013-108', 'PUH-2016-064', 'PUH-2018-068', 'PUH-2015-074', 'PUH-2016-125', 'PUH-2010-080', 'PUH-2010-087', 'PUH-2017-008', 'PUH-2019-033', 'PUH-2016-154', 'PUH-2017-222', 'PUH-2010-163', 'PUH-2015-209', 'PUH-2017-146', 'PUH-2013-119', 'PUH-2019-010', 'PUH-2012-153', 'PUH-2015-064', 'PUH-2014-100', 'PUH-2016-087', 'PUH-2012-172', 'PUH-2015-216']\n",
    "\n",
    "save_path1 = '/heinz-georgenas/EEG/0_6hr_60_avg/'\n",
    "save_path2 = '/heinz-georgenas/EEG/6_12hr_60_avg/'\n",
    "\n",
    "saved_files2 = glob.glob(save_path2 + '*.csv')\n",
    "saved_patients2 = [file.split('/')[-1] for file in saved_files2]\n",
    "patients_with_more_than_6hr = sorted([file.split('/')[-1][:len('PUH-xxxx-xxx')] for file in saved_files2])\n",
    "\n",
    "time_to_outcome_path = '/heinz-georgenas/users/xiaobins/eeg/data/Time to outcome.csv'\n",
    "time_to_outcome = pd.read_csv(time_to_outcome_path)\n",
    "time_to_outcome.dropna(inplace=True)\n",
    "time_to_outcome['awake'] = time_to_outcome.follow_com.apply(lambda x: 1 if x=='Yes' else 0)\n",
    "\n",
    "te_all_eeg = []\n",
    "\n",
    "for patient in tqdm(te_id_1234):\n",
    "    first_6hr = pd.read_csv(save_path1+patient+'-12hours.csv', header=None)\n",
    "    if patient in patients_with_more_than_6hr:\n",
    "        after_6hr = pd.read_csv(save_path2+patient+'-12hours.csv', header=None)\n",
    "        curr_X = pd.concat([first_6hr, after_6hr], axis=0)\n",
    "        curr_X = np.array(curr_X)[:, 2550:2581]\n",
    "        curr_X = np.concatenate((curr_X[:, :2], curr_X[:, -10:]), axis=1)\n",
    "        curr_y = np.array(time_to_outcome[time_to_outcome.id==patient][['obsduration', 'awake']]).flatten()\n",
    "        time = math.floor(curr_y[0] * 60)\n",
    "        curr_X = curr_X[:time] # do NOT use EEG that are later than time-to-outcome!\n",
    "    te_all_eeg.append(curr_X)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0137940a",
   "metadata": {},
   "outputs": [],
   "source": [
    "## when running for the first time, save the test data to a file for reuse to save time\n",
    "import pickle\n",
    "with open('high_res_combined_data_test.pickle', 'wb') as handle:\n",
    "    pickle.dump(te_all_eeg,\n",
    "                handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7db43f53",
   "metadata": {
    "tags": []
   },
   "source": [
    "### Set Hyper-Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "4a247a42-cf90-4385-9dc6-86833df3b279",
   "metadata": {
    "collapsed": true,
    "execution": {
     "iopub.execute_input": "2023-04-24T07:24:21.376577Z",
     "iopub.status.busy": "2023-04-24T07:24:21.376338Z",
     "iopub.status.idle": "2023-04-24T07:24:21.386375Z",
     "shell.execute_reply": "2023-04-24T07:24:21.385809Z",
     "shell.execute_reply.started": "2023-04-24T07:24:21.376556Z"
    },
    "jupyter": {
     "outputs_hidden": true
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "24\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[{'beta': 0.5, 'dropout': 0.2, 'gamma': 0.1, 'lr_train': 0.0001},\n",
       " {'beta': 0.5, 'dropout': 0.2, 'gamma': 0.1, 'lr_train': 0.0005},\n",
       " {'beta': 0.5, 'dropout': 0.2, 'gamma': 0.1, 'lr_train': 0.001},\n",
       " {'beta': 0.5, 'dropout': 0.2, 'gamma': 0.5, 'lr_train': 0.0001},\n",
       " {'beta': 0.5, 'dropout': 0.2, 'gamma': 0.5, 'lr_train': 0.0005},\n",
       " {'beta': 0.5, 'dropout': 0.2, 'gamma': 0.5, 'lr_train': 0.001},\n",
       " {'beta': 0.5, 'dropout': 0.3, 'gamma': 0.1, 'lr_train': 0.0001},\n",
       " {'beta': 0.5, 'dropout': 0.3, 'gamma': 0.1, 'lr_train': 0.0005},\n",
       " {'beta': 0.5, 'dropout': 0.3, 'gamma': 0.1, 'lr_train': 0.001},\n",
       " {'beta': 0.5, 'dropout': 0.3, 'gamma': 0.5, 'lr_train': 0.0001},\n",
       " {'beta': 0.5, 'dropout': 0.3, 'gamma': 0.5, 'lr_train': 0.0005},\n",
       " {'beta': 0.5, 'dropout': 0.3, 'gamma': 0.5, 'lr_train': 0.001},\n",
       " {'beta': 1, 'dropout': 0.2, 'gamma': 0.1, 'lr_train': 0.0001},\n",
       " {'beta': 1, 'dropout': 0.2, 'gamma': 0.1, 'lr_train': 0.0005},\n",
       " {'beta': 1, 'dropout': 0.2, 'gamma': 0.1, 'lr_train': 0.001},\n",
       " {'beta': 1, 'dropout': 0.2, 'gamma': 0.5, 'lr_train': 0.0001},\n",
       " {'beta': 1, 'dropout': 0.2, 'gamma': 0.5, 'lr_train': 0.0005},\n",
       " {'beta': 1, 'dropout': 0.2, 'gamma': 0.5, 'lr_train': 0.001},\n",
       " {'beta': 1, 'dropout': 0.3, 'gamma': 0.1, 'lr_train': 0.0001},\n",
       " {'beta': 1, 'dropout': 0.3, 'gamma': 0.1, 'lr_train': 0.0005},\n",
       " {'beta': 1, 'dropout': 0.3, 'gamma': 0.1, 'lr_train': 0.001},\n",
       " {'beta': 1, 'dropout': 0.3, 'gamma': 0.5, 'lr_train': 0.0001},\n",
       " {'beta': 1, 'dropout': 0.3, 'gamma': 0.5, 'lr_train': 0.0005},\n",
       " {'beta': 1, 'dropout': 0.3, 'gamma': 0.5, 'lr_train': 0.001}]"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import ParameterGrid\n",
    "param_grid = {'dropout': [0.2, 0.3],#, 1e-4], \n",
    "              'lr_train': [1e-4, 5e-4, 1e-3],#, 1e-4],\n",
    "              'beta': [0.5, 1],#, 1e-3],\n",
    "              'gamma': [0.1, 0.5],\n",
    "             }\n",
    "param_grid_list = list(ParameterGrid(param_grid))\n",
    "print(len(param_grid_list))\n",
    "param_grid_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "3e7374c9",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2023-03-03T18:15:31.543716Z",
     "start_time": "2023-03-03T18:15:31.531758Z"
    },
    "execution": {
     "iopub.execute_input": "2023-04-24T07:24:29.698710Z",
     "iopub.status.busy": "2023-04-24T07:24:29.698466Z",
     "iopub.status.idle": "2023-04-24T07:24:29.728337Z",
     "shell.execute_reply": "2023-04-24T07:24:29.727804Z",
     "shell.execute_reply.started": "2023-04-24T07:24:29.698689Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "cuda\n"
     ]
    }
   ],
   "source": [
    "boost_mode = 'ON' #{'ON', 'OFF'}\n",
    "\n",
    "param = {'batch_size': 32,\n",
    "\n",
    "         'num_epoch_burn_in': 20,\n",
    "         'num_epoch': 100,\n",
    "\n",
    "         'dropout': 0.2, # 1 - keep_prob\n",
    "         'lr_train': 1e-4,\n",
    "\n",
    "         'hidden_rnn': 64,\n",
    "         'hidden_dim_FC': 64,\n",
    "         'layers_rnn': 2,\n",
    "         'hidden_att': 2,\n",
    "         'hidden_cs' : 2,\n",
    "\n",
    "         'alpha' :1.0,\n",
    "         'beta'  :1,\n",
    "         'gamma' :0.1\n",
    "        }\n",
    "\n",
    "\n",
    "batch_size = param['batch_size']\n",
    "num_epoch  = param['num_epoch']\n",
    "\n",
    "dropout    = param['dropout']\n",
    "lr_train   = param['lr_train']\n",
    "\n",
    "hidden_rnn = param['hidden_rnn']\n",
    "hidden_dim_FC = param['hidden_dim_FC']\n",
    "layers_rnn = param['layers_rnn']\n",
    "hidden_att = param['hidden_att']\n",
    "hidden_cs = param['hidden_cs']\n",
    "\n",
    "alpha      = param['alpha']\n",
    "beta       = param['beta']\n",
    "gamma      = param['gamma']\n",
    "\n",
    "device     = torch.device('cuda' if torch.cuda.is_available() else 'cpu')\n",
    "print(device)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bc506fb7",
   "metadata": {},
   "source": [
    "This implementation does not consider missing values."
   ]
  },
  {
   "cell_type": "markdown",
   "id": "60f3877b-f0d1-4ea5-b505-bb6e96d61907",
   "metadata": {},
   "source": [
    "# Part II: Error bars for CV ROC surve"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4a51ddf5-a67f-48b3-94f2-cd559496330d",
   "metadata": {},
   "outputs": [],
   "source": [
    "'ddrsa_eeg_cr3_1234_alpha1.0_beta5_gamma0.5_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230318_0133',\n",
    "'ddrsa_eeg_cr3_1111_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230324_0049',\n",
    "'ddrsa_eeg_cr3_2222_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230317_0946',\n",
    "'ddrsa_eeg_cr3_3333_alpha1.0_beta1_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230328_0242',\n",
    "'ddrsa_eeg_cr3_4444_alpha1.0_beta5_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230328_0907',\n",
    "\n",
    "best_pt_path = 'ddrsa_eeg_cr3_1234_sigmoidFalse_alpha1.0_beta0.5_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt3_hiddenCS3_dropout0.4_20230323_1331'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "a6b52898-c902-4ebb-ba38-fdcfda09d3a6",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-24T07:32:10.053471Z",
     "iopub.status.busy": "2023-04-24T07:32:10.053220Z",
     "iopub.status.idle": "2023-04-24T07:32:10.668195Z",
     "shell.execute_reply": "2023-04-24T07:32:10.667374Z",
     "shell.execute_reply.started": "2023-04-24T07:32:10.053449Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns; sns.set()\n",
    "plt.style.use('default')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "a2ed88fb-5a4a-45f9-856c-300bd175021e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-24T07:30:02.024358Z",
     "iopub.status.busy": "2023-04-24T07:30:02.024110Z",
     "iopub.status.idle": "2023-04-24T07:30:02.053876Z",
     "shell.execute_reply": "2023-04-24T07:30:02.053306Z",
     "shell.execute_reply.started": "2023-04-24T07:30:02.024336Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_Event"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "da466772-5424-4f25-936f-e2929b90978c",
   "metadata": {},
   "source": [
    "### Hour 6"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "ef54183e-094d-4c90-a0b6-d446a8301c90",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-24T07:32:58.428273Z",
     "iopub.status.busy": "2023-04-24T07:32:58.428003Z",
     "iopub.status.idle": "2023-04-24T07:33:20.953527Z",
     "shell.execute_reply": "2023-04-24T07:33:20.952782Z",
     "shell.execute_reply.started": "2023-04-24T07:32:58.428252Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/heinz-georgenas/anaconda3/envs/py39xiaobin/lib/python3.9/site-packages/pycox/preprocessing/label_transforms.py:113: UserWarning: Calling fit method, when 'cuts' are already defined. Leaving cuts unchanged.\n",
      "  warnings.warn(\"Calling fit method, when 'cuts' are already defined. Leaving cuts unchanged.\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eval time: 24 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5it [00:06,  1.32s/it]\n",
      "/tmp/ipykernel_2964573/668318695.py:139: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\n",
      "Invalid limit will be ignored.\n",
      "  ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DDRSA (AUC = 0.868 ± 0.031)\n",
      "eval time: 48 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5it [00:05,  1.19s/it]\n",
      "/tmp/ipykernel_2964573/668318695.py:139: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\n",
      "Invalid limit will be ignored.\n",
      "  ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DDRSA (AUC = 0.863 ± 0.028)\n",
      "eval time: 72 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5it [00:05,  1.19s/it]\n",
      "/tmp/ipykernel_2964573/668318695.py:139: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\n",
      "Invalid limit will be ignored.\n",
      "  ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DDRSA (AUC = 0.849 ± 0.031)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn import metrics\n",
    "from tqdm import tqdm\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns; sns.set()\n",
    "plt.style.use('default')\n",
    "import matplotlib\n",
    "import pickle\n",
    "\n",
    "\n",
    "use_sigmoid = True\n",
    "\n",
    "seed_list = [1234, 1111, 2222, 3333, 4444]\n",
    "best_pt_path_list = ['ddrsa_eeg_cr3_1234_alpha1.0_beta5_gamma0.5_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230318_0133',\n",
    "                     'ddrsa_eeg_cr3_1111_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230324_0049',\n",
    "                     'ddrsa_eeg_cr3_2222_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230317_0946',\n",
    "                     'ddrsa_eeg_cr3_3333_alpha1.0_beta1_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230328_0242',\n",
    "                     'ddrsa_eeg_cr3_4444_alpha1.0_beta5_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230328_0907',\n",
    "                    ]\n",
    "\n",
    "data_mode = 'EEG_cr3' # ['EEG_nocr', 'EEG_cr2', 'EEG_cr4']\n",
    "\n",
    "if data_mode == 'EEG_cr3':\n",
    "    path = '../../eeg/competing-risk/EEG_processed_data_long_by_death_cat_exclude_3.csv'\n",
    "\n",
    "(x_dim, x_dim_cont, x_dim_bin), (data, time, label, time_original, time_to_last), \\\n",
    "    (mask1, mask2, mask3), (data_mi), trans_discrete_time = \\\n",
    "                impt.import_dataset(path)\n",
    "\n",
    "_, num_Event, num_Category  = np.shape(mask1)  # dim of mask3: [subj, Num_Event, Num_Category]\n",
    "max_length                  = np.shape(data)[1]\n",
    "\n",
    "pred_time = list(range(1,13))\n",
    "eval_time = [24, 48, 72]\n",
    "mean_fpr = np.linspace(1e-10, 1, 1000)\n",
    "\n",
    "tprs = []\n",
    "aucs = []\n",
    "\n",
    "for e, e_time in enumerate(eval_time):\n",
    "    print('eval time:', e_time, end=' ')\n",
    "    tprs = []\n",
    "    aucs = []\n",
    "    for r, seed in tqdm(enumerate(seed_list)):\n",
    "\n",
    "        best_pt_path = best_pt_path_list[r]\n",
    "\n",
    "        ### TRAINING-TESTING SPLIT\n",
    "        (tr_data,te_data, tr_data_mi, te_data_mi, tr_time,te_time, tr_label,te_label, tr_time_orinal,te_time_original,\n",
    "         tr_time_to_last,te_time_to_last,\n",
    "         tr_mask1,te_mask1, tr_mask2,te_mask2, tr_mask3,te_mask3) = \\\n",
    "                train_test_split(data, data_mi, time, label, time_original, time_to_last,\n",
    "                                 mask1, mask2, mask3, test_size=0.2, stratify=label, random_state=seed) \n",
    "\n",
    "        (tr_data,va_data, tr_data_mi, va_data_mi, tr_time,va_time, tr_label,va_label, tr_time_orinal,va_time_original,\n",
    "         tr_time_to_last, va_time_to_last,\n",
    "         tr_mask1,va_mask1, tr_mask2,va_mask2, tr_mask3,va_mask3) = \\\n",
    "                train_test_split(tr_data, tr_data_mi, tr_time, tr_label, tr_time_orinal, tr_time_to_last,\n",
    "                                 tr_mask1, tr_mask2, tr_mask3, test_size=0.2, stratify=tr_label, random_state=seed) \n",
    "\n",
    "        model_name = 'ddrsa_eeg_cr3'\n",
    "        num_Event = 3\n",
    "\n",
    "        model = \\\n",
    "            DDRSA(input_dim  = x_dim,\n",
    "                  output_dim = num_Category,\n",
    "                  layers_rnn = param['layers_rnn'],\n",
    "                  hidden_rnn = param['hidden_rnn'],\n",
    "                  long_param = {'layers': 2*[param['hidden_dim_FC']], 'dropout': dropout}, \n",
    "                  att_param  = {'layers': param['hidden_att']*[param['hidden_dim_FC']], 'dropout': dropout}, \n",
    "                  cs_param   = {'layers': param['hidden_cs']*[param['hidden_dim_FC']], 'dropout': dropout},\n",
    "                  risks      = num_Event,\n",
    "                  use_sigmoid = use_sigmoid\n",
    "                 )\n",
    "        optimizer = torch.optim.Adam(model.parameters(), lr = lr_train)\n",
    "        _ = load_checkpoint(f'checkpoints_ddrsa/{model_name}/{best_pt_path}_{seed}.pt', model, optimizer, device)\n",
    "        model = model.to(device)\n",
    "\n",
    "        prob_death_all = np.ones((te_data.shape[0], len(pred_time), len(eval_time)))\n",
    "        prob_awake_all = np.ones((te_data.shape[0], len(pred_time), len(eval_time)))\n",
    "        risk_death_old_all = np.ones((te_data.shape[0], len(pred_time), len(eval_time)))\n",
    "\n",
    "        for p, p_time in enumerate(pred_time):\n",
    "            pred_horizon = int(p_time)\n",
    "            out = _f_get_pred(model, te_data, p_time, device)\n",
    "            risk_death_awaken_all_future = 1 - out[:, :2, :pred_horizon].sum((1,2))\n",
    "\n",
    "            for t, t_time in enumerate(eval_time):\n",
    "                eval_horizon = int(t_time) + pred_horizon\n",
    "                risk_death = out[:, 0, pred_horizon:eval_horizon+1].sum(1)\n",
    "                risk_awake = out[:, 1, pred_horizon:eval_horizon+1].sum(1)\n",
    "                risk_death_old = risk_death / (risk_death+risk_awake)\n",
    "                prob_death = risk_death / risk_death_awaken_all_future\n",
    "                prob_awake = risk_awake / risk_death_awaken_all_future\n",
    "                prob_death_all[:, p, t] = prob_death\n",
    "                prob_awake_all[:, p, t] = prob_awake\n",
    "                risk_death_old_all[:, p, t] = risk_death_old\n",
    "\n",
    "        mask = (te_label==1) + (te_label==2) # for only people who died naturally, and who awoke\n",
    "\n",
    "        scores = -(prob_awake_all / prob_death_all)[:, 5, e]\n",
    "        fpr, tpr, thresholds = metrics.roc_curve(te_label[mask], scores[mask.flatten()], \n",
    "                                                 pos_label=1)\n",
    "        interp_tpr = np.interp(mean_fpr, fpr, tpr)\n",
    "        interp_tpr[0] = 0.0\n",
    "        roc_auc = metrics.auc(fpr, tpr)\n",
    "\n",
    "        tprs.append(interp_tpr)\n",
    "        aucs.append(roc_auc)\n",
    "\n",
    "    n_std = 1\n",
    "    scale = 'log' # {'log', 'linear'}\n",
    "\n",
    "    fig = plt.figure(figsize=(12,10))\n",
    "    ax = fig.add_subplot(3, 2, 1)\n",
    "    ax.plot(mean_fpr, mean_fpr, linestyle='--', color='black', alpha=.8,\n",
    "            label=r'%s (AUC = %0.3f $\\pm$ %0.3f)' % ('Random', 0, 0))\n",
    "\n",
    "    mean_tpr = np.mean(tprs, axis=0)\n",
    "    mean_tpr[-1] = 1.0\n",
    "    mean_auc = metrics.auc(mean_fpr, mean_tpr)\n",
    "    std_auc = np.std(aucs)\n",
    "    plt.plot(mean_fpr, mean_tpr, color='b',\n",
    "            label=r'%s (AUC = %0.3f $\\pm$ %0.3f)' % ('DDRSA', mean_auc, n_std*std_auc),\n",
    "            lw=2, alpha=.8)\n",
    "    print(r'%s (AUC = %0.3f ± %0.3f)' % ('DDRSA', mean_auc, n_std*std_auc))\n",
    "    std_tpr = np.std(tprs, axis=0)\n",
    "    tprs_upper = np.minimum(mean_tpr + n_std*std_tpr, 1)\n",
    "    tprs_lower = np.maximum(mean_tpr - n_std*std_tpr, 0)\n",
    "    ax.fill_between(mean_fpr, tprs_lower, tprs_upper, color='b', alpha=.2)\n",
    "\n",
    "    plt.xlim([2e-2, 5])\n",
    "    plt.yticks([0.00,0.25,0.50,0.75,1.00])\n",
    "    if scale == 'log':\n",
    "        ax.set_xscale(scale)\n",
    "        ax.set_xticks([0.03, 0.06, 0.10, 0.18, 0.32, 0.56, 1.00])\n",
    "        ax.get_xaxis().set_major_formatter(matplotlib.ticker.ScalarFormatter())\n",
    "\n",
    "    ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n",
    "           # title=r'Mean ROC with $\\pm$ %0.1f std. dev.' % (n_std)\n",
    "          )\n",
    "    ax.legend(loc='lower right')\n",
    "    plt.xlabel('False positive rate')\n",
    "    plt.ylabel('True positive rate')\n",
    "    \n",
    "    auc_info = {'mean_tpr': mean_tpr, 'mean_auc': mean_auc, 'std_auc': std_auc, 'tprs': tprs}\n",
    "    with open(f'ddrsa_auc_info_{6}_{e_time}.pickle', 'wb') as handle:\n",
    "        pickle.dump(auc_info,\n",
    "                    handle, protocol=pickle.HIGHEST_PROTOCOL)\n",
    "    \n",
    "    fname = r'ROC_%s_%d_%d.pdf' % ('DDRSA', 6, e_time)\n",
    "    plt.savefig(fname, dpi=200, format='pdf', bbox_inches='tight')\n",
    "    # plt.show()\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "b8d5f202-d2aa-4c01-a113-9247ec833e4d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-24T07:35:30.014973Z",
     "iopub.status.busy": "2023-04-24T07:35:30.014549Z",
     "iopub.status.idle": "2023-04-24T07:35:54.954607Z",
     "shell.execute_reply": "2023-04-24T07:35:54.954017Z",
     "shell.execute_reply.started": "2023-04-24T07:35:30.014941Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/heinz-georgenas/anaconda3/envs/py39xiaobin/lib/python3.9/site-packages/pycox/preprocessing/label_transforms.py:113: UserWarning: Calling fit method, when 'cuts' are already defined. Leaving cuts unchanged.\n",
      "  warnings.warn(\"Calling fit method, when 'cuts' are already defined. Leaving cuts unchanged.\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "eval time: 24 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5it [00:06,  1.32s/it]\n",
      "/tmp/ipykernel_2964573/3238421982.py:139: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\n",
      "Invalid limit will be ignored.\n",
      "  ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DDRSA (AUC = 0.871 ± 0.015)\n",
      "eval time: 48 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5it [00:08,  1.71s/it]\n",
      "/tmp/ipykernel_2964573/3238421982.py:139: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\n",
      "Invalid limit will be ignored.\n",
      "  ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DDRSA (AUC = 0.853 ± 0.019)\n",
      "eval time: 72 "
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "5it [00:05,  1.17s/it]\n",
      "/tmp/ipykernel_2964573/3238421982.py:139: UserWarning: Attempted to set non-positive left xlim on a log-scaled axis.\n",
      "Invalid limit will be ignored.\n",
      "  ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DDRSA (AUC = 0.822 ± 0.020)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1200x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAgEAAAEhCAYAAADmonPKAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjQuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/MnkTPAAAACXBIWXMAAA9hAAAPYQGoP6dpAABsr0lEQVR4nO3deXxM1/sH8M/MZJbsq6yC2GKvJIqgVIug9mqjtZZWtfZUW1VqF+23domidoqidEOpXVHEVqKxhSCJyDpJJrPe8/sjv9xmZJLMJJOZLM/79ZoXc+fcO88ZZu5zzz2LgDHGQAghhJAaR2jtAAghhBBiHZQEEEIIITUUJQGEEEJIDUVJACGEEFJDURJACCGE1FCUBBBCCCE1FCUBhBBCSA1lY+0AKiOO45CYmAhHR0cIBAJrh0MIIYSAMYbs7Gz4+vpCKDTPNTwlAQYkJibC39/f2mEQQgghRTx+/Bi1a9c2y7EoCTDA0dERQP4H7eTkZOVoCCGEEEAul8Pf358/R5kDJQEGFNwCcHJyoiSAEEJIpWLO29TUMZAQQgipoSgJIIQQQmooSgIIIYSQGoqSAEIIIaSGsmoScPr0afTt2xe+vr4QCAQ4cOBAqfucOnUKISEhkMlkqF+/Pr777rsiZfbt24dmzZpBKpWiWbNm2L9/fwVETwghhFRtVk0CcnNz8dJLL2H16tVGlY+Pj0fv3r3xyiuv4OrVq5gxYwYmTZqEffv28WXOnz+P8PBwDB8+HNevX8fw4cPx9ttv4++//66oahBCCCFVkoAxxqwdBJA/5GH//v0YMGBAsWU+//xz/PLLL7h9+za/bdy4cbh+/TrOnz8PAAgPD4dcLsehQ4f4Mj179oSrqyt27txpVCxyuRzOzs7IysqiIYKEEEIqnI5jUORpsPvYHSQ9zwEEgEgoxOi+zeHtbg+gYs5NVWqegPPnz6NHjx5628LCwrBhwwZoNBqIxWKcP38eU6dOLVJm+fLlxR5XpVJBpVLxz+VyuVnjJoQQQgpwHENCshxqHQchBHCwEyMzW4WYf5/h0q1kKHJzoMjLhYe7B3RcxV6nV6mOgcnJyfDy8tLb5uXlBa1Wi9TU1BLLJCcnF3vcyMhIODs78w+aMpgQQkhF0HEMSrUW6XIlUjPy8Cw9F3cTMvHH349w+upT6DgGtVqFHLkc6rxsVPTyNVWqJQAoOlNSwd2MwtsNlSlphqUvvvgCERER/POCqRkJIYQQc1FrdLj7OBMarQ4qjQ52UhvcfJCGv24kAgCEQgE4xuDp6YmIIS3RqmkAbG1tKzSmKpUEeHt7F7miT0lJgY2NDdzd3Uss82LrQGFSqRRSqdT8ARNCCKnxlCot0uR5UKp0yM3TQKfjcP9pFn/yf1F9P2e0DWpqkVVsq9TtgNDQUBw9elRv25EjR9CmTRuIxeISy3To0MFicRJCCCEFFCotHibK8TwjDwIAF2OTiyQAjAGcOgddWzlhZG/LJACAlZOAnJwcXLt2DdeuXQOQPwTw2rVrSEhIAJDfTD9ixAi+/Lhx4/Do0SNERETg9u3b2LhxIzZs2IBp06bxZSZPnowjR47g66+/xr///ouvv/4af/75J6ZMmWLJqhFCCKmhdDoOao2OfyjVWgiFAni42MLVSYbnmcr/CrP8UQCvBXvi6sFvsXjmx7gTd7v4g5uZVW8HXL58GV27duWfF9yXHzlyJDZv3oykpCQ+IQCAgIAAHDx4EFOnTkVUVBR8fX2xcuVKvPnmm3yZDh06YNeuXZg5cyZmzZqFBg0aYPfu3WjXrp3lKkYIIaRGYozh7uNMKJSaF17Jv9//LE0BlVoLAJCKRXiltR/q+Dhg/+6t0KgUCGzUAG3btLZYvJVmnoDKhOYJIIQQUhKlSmtw+J6OY4hPzIJGw8FW+t91to1IgCMXExAbn8Zvc7KXYECXBnCRaTB6xDvQaNRYsmQJunTpYvA9a/w8AYQQQoi1KFVaKFRaaLQcUtIVyPv/K/oXabQ6uDrIIBGLAOS3DuSptHjwNIsvo9VxsJPZQCQUYvvWjdBo1AgKCkLnzp0tUpcClAQQQgghL0hJz0WeSv8kn6fSIi1LmT+Uj2Nwc7KF0ED/PYb/hvuBAXuO38WTlGz+dTuZGE3ruSG0hTdyMpLwx+GDAPL7tFmqQ2ABSgIIIYSQF6Rk5OF5Zh7sZWK97S6OMkj//wrfkPjELBw6/7BIAlGYr4c9Wjeuhfq1XfD5inkAY+jWrRtatGhhtviNRUkAIYQQYoCrgxQOdpIi2+8+zsCpq0+h1uiKvFbSyR8AWjbwwEuNPSAAwDgOrVu3xq1btzB+/HhzhW0SSgIIIYSQF3AMKDxnr1qrw9G/E5CSoUC6XFn8joX4ejjwf3d1kuLV4NqQikVQazgoNVoIRUKMHj0aQ4cOtdqEdZQEEEIIIYVwHAMY05u3/97jTPz7KL1IWVfHoidvG5EQLzf1QtMAd73tWi2HZxkKiEVC2MrEEP7/G1hzxlpKAgghhNRYeSotnqZkgyu0jXH5vffFIpFeuQI2IiHsZDZ4Nbg2Gvm7Gv1eOo7BTmYDX3dbLJz3FYYNfcfqc9hQEkAIIaTGUql1eJ6ZB4nNC539BIBU8t82re6/NKFPxwDU9XECxxjU2qL9AlRqHV6cgUcsEoJjDEKBAId/34/z587i3t04/Pzzz5BIivY7sBRKAgghhNRYBfPlOTv81ySv1XK4/Sgd2blqftvjlBz+7wIBkJaVB5nE8ClUJBJAIhZB+P/jBzkuf54AHcegU+Vg48aNAICPPvrIqgkAQEkAIYSQGuz/F6PX23YlLgVnrj8tUpbjGDjGIFdo4Olmh/p+LpDYGFiCRwC9BEH3/0kAACz93/dQ5OYiMDAQffr0MWNNyoaSAEIIIdWWPFcNHccV+3pmtgoM+m33zzMVBstyjMHRToL2LbxhK7WBk71xV/EioQAOtmLcuXMHv/32CwBg2rRpEAqtv5AvJQGEEEKqNLVGh5R0xf8P6wNEwvyV+bQ6DmlZSuTkqfme+C/SajmkZyuhVP13b1+e+9/iP3061ecnB8rKVaFNEy/U9nQ0OUbGGJYsWQKO49C9e3cEBQWZfIyKQEkAIYSQKi1PpcXjlGz+RM/Yf9f2AoEAtVzsDCYBHGPYfug2nmfmFXvsAF8nvtPg88w8yKRlO21evnwZMTExkEgkmDRpUpmOUREoCSCEEFIlqDU6gyv35eblX7m7O9sa3Ke4WfxyFJoSEwAXBynEov+a7AVgKOvM/m3atMHXX3+N1NRU+Pj4lPEo5kdJACGE1HBaHcefSC1NKMzvSV/ayZXjGO4/zSp1Wl7gvx7/1++l4sTlx/kL+ZTCy80O9Xz+W55XKBSgSR03vQV9GPBiH0KjCQQCvP7662XbuQJREkAIITVcZrYKD55mldiBrqKIhELYiIw7s6o0OjjIJLAx0CNfJMzvpHf+nyT8fSvJYItBSZrUdUObpl6llBJAYGIWkJWVBaFQCEdH0/sRWAIlAYQQUsNxHAPHcajlYmfx99ZqOaNP2DKpTdFJfQBcuJmEa3eeQ6PjDC7qAwAN/JyLPa6rowwtG3qU+v4C6C0nYJQVK1bg9OnTmDVrFrp06WLazhZASQAhhNRwHHtxkJzl2NgIy3UiUmt0OP9PksEmf18PB0glIrRv7g3fWg4G9i5eujwPOo69cDuAmZQE3LhxA7/8kj8k0M3NzaT3txRKAgghpIbjGDO5mbuyUGl0fAIgthHCwVYMe1sJXn/ZHx7OtlCqteA4BoWy+D4PAoFAb/QAA4OOY/DxcICDrbhQOcDRwNLChuh0OixevBgA0K9fP7Rs2bIs1atwlAQQQkgNp9OZdoVbmWi1//VjaOTvgl6hAf+9puOQrVDDTiY2tCtPp+OKTBjkYCuBp6ttqfsWZ9++fbhz5w4cHR0xceLEMh3DEigJIISQGk7HcXrN3pVRbp4G1+4+LzI6QPnC6n5qrQ7pciVE/z9vv71MjMZ1XSE20JegAMdxRRb8EQgEEBuaEtgI6enpiI6OBgCMHz8erq7GrzRoaZQEEEJIDfIwMQtyhX7TOMdxxc6oV5HUGh1uxachT1n6sL/zN5NKLSO2EUGr5eBoJ0EdL0cIhQIIhYJiF/opIBIWnyCUxYoVK5CTk4OmTZti0KBBZj22uVESQAghVZhOx+Hp81zodKUP79MxBnmOCmotp3diTM9S4Xmmgl/1zlJOxjwxagx/qVj+fXx3ZxnSspTwcLGFq5Os/MctA47jIBaLIRQKMX369EqxPkBJBIyZ41+gepHL5XB2dkZWVhacnJxK34EQQswkLSsPGXKl0eU1Wg5ZOflz4xtzMS8QCODkIOGv/LMVamz45abJ4+qtxdPVDmHt6upty1Vp4WwvQW0vBwgggEQsNDh7oCUlJibC19fXrMesiHMTtQQQQoiFMcag0uhgaFzes3QFnmfkwd7W+A5ptlIbk8q/+H7WTgBsREL079yg1HISGyG8PeyL3LrIUahhYyOEr4dpwwArkrkTgIpCSQAhhFiYPFeN+MQs6HRFT74aLQc3Jxlsy7hQTWnyVFrkFhoul1po7vyWDTxQ29OyJ1KxjRB1fZwMTgJkLMZMn8TH3FJSUrBkyRJMnDgRtWvXtm4wJqAkgBBCLEyr45Cj0MDNueh9azuUvVd6aeITs/Dz6fvFXvkH+DqhkX/l7cleHIb86Yet6ZtvvsHJkyeRmZmJtWvXWjUWU1ASQAghFqbTMYhEgnJd/ZbFnYSMEpv+Pax8H72sGLPuPAcnTpzAyZMnIRKJ8Omnn1ovkDKgJIAQQixMx1mn/Vqp/m9e/ab13GBTaJncuj5OVutRb6xshdrgLZRshRpODsbN5GduOTk5+OabbwAAI0eORMOGDa0SR1lREkAIIWUgz1UjK0dV7Jz7Go0OGq3hYXtaHQeRVZKA/8bj92hXVy8JsBaFUmPwxP4iLceB4xic7CUQi/VbUFwcJXCyt04SEBUVhefPn6NOnTp4//33rRJDeVASQAghZZCVo8K9x5mwkxn+GRUKBBCWsESusXPQlyY+MQvn/kkqNuEoLCtHBSC/N35lSAAYY8hRauBsL0XpUxSIILYRoZ6vU6WIHchfIGjv3r0AgBkzZkAisU4iUh6UBBBCSBkoVBo42Uvg7CA12zG1Og6nrj5Baqbx8wQ8Sck2+X0MJS6cFYYJ6jgGG5EQ9XycyjzE0Zq2b98Oxhj69euHNm3aWDucMqEkgBBSo6g0OmRmq8p1DMYYVCqd2a9I/32Yjmt3npd5f4nYuI6Gzeu743mhoYEAjLgSrxi2UhuISmgxqcwWLFiA7du3480337R2KGVGSQAhpNLS6ThozXyFmpKuwMMkuRlO4AyujmXrSPf4WTbuPs7EixO2Jqbmlul4tlIb9HulPmp7OpZa9nmmAr4eDkU60tmIhFZZTNiYuf0rK4lEgtGjR1s7jHKpmp88IaRGeJScjYxs45vGjaHjGBxkYjhaqSOZUqXF/lP3Sr2H/+HAVpBJjLuyFwoFRi8AJADg4ig1622MmoTjOPz666/o06cPRCLLDvGsCJQEEEIsQqnWFlkGtiRqDYeMbCWEEEBq5MnQWBU1GU8BnY7DvSeZyFPpiryWlasqNQGo7+sMhwq4R57f8iCo9MsGV2Y//PADli9fjsOHDyM6OrrKf5aUBBBCLCItKw8Pk+QmL1lby8WugiIqn8wcFZ6lKwy+dvbaU2TmlN7vILSlDxr4uehtE4kEcK+g8foF0+ta6/5/Vffo0SNER0cDAHr27FnlEwCgjElAZmYm9u7di/v37+PTTz+Fm5sbrly5Ai8vL/j5+Zk7RkJINaDjABuh9Vd3M4d0uRJbfo8t1zK4NiIhmtd3h7O95Zrl82fWo5aAsuA4DnPnzoVarUZoaCj69etn7ZDMwuQk4MaNG+jWrRucnZ3x8OFDfPDBB3Bzc8P+/fvx6NEjbN26tSLiJIRUcRzHVZuTT8KzbKMSAJnEBl1DDC8m4+Nub9EEAChoCTBuyWGib+fOnbhx4wbs7Owwc+bMavN/2eQkICIiAqNGjcI333wDR8f/eqL26tUL7777rlmDI4RUHzqdded3N6dcxX+r8L3UsBZcnYqezGUSGzSu41rh/Q+A/KV0jRlEodHmD2s09ZZMTZeQkICoqCgA+edALy8vK0dkPib/77x06RI+/PDDItv9/PyQnJxscgDR0dEICAiATCZDSEgIzpw5U2zZUaNG8U1ZhR/Nmzfny2zevNlgGaXSvD2MCSGm0f1/U3R1kJOn5v/+UiMPhDTxKvJoXt+9TAmAjmNQqrVGP7JyVdDoGGQSEexkNiU+CiY3srFAYlKdLF68GGq1Gu3bt0f//v2tHY5ZmdwSIJPJIJfLi2yPi4tDrVq1TDrW7t27MWXKFERHR6Njx45Yu3YtevXqhdjYWNSpU6dI+RUrVmDx4sX8c61Wi5deeglvvfWWXjknJyfExcUViZsQYj1cJWgJYIzh4Ll43EnILNf9/MIcbM071DBDngeJCaMhBIL8IX+N/F3MGgf5z2effYZvv/22Wt0GKGByEtC/f3/MmzcPP/74I4D8+0sJCQmYPn26ybMmLV26FGPGjOEXXVi+fDn++OMPrFmzBpGRkUXKOzs7w9nZmX9+4MABZGRk4L333tMrJxAI4O3tbWrVCCEVSMdxVm+GTk5X4N9HGWY7nkxiA5nUvMMXOQC+Hg5wdzb+wkUopCv7ilSvXj2sXr3a2mFUCJOTgG+//Ra9e/eGp6cn8vLy0KVLFyQnJyM0NBQLFy40+jhqtRoxMTGYPn263vYePXrg3LlzRh1jw4YN6NatG+rWrau3PScnB3Xr1oVOp0Pr1q0xf/58BAUFFXsclUoFleq/4TyGWjoIIeXDcTDpKipPqcUffz9EZo669MJGUhVaRc/VUWb0ZDyGiERCBAfWKrZOOXkaaHXFzAfAAGZo/UGWP5mPrdQGYpuqPxFNVaZWqxEXF4eWLVtaO5QKZXIS4OTkhLNnz+L48eO4cuUKOI5DcHAwunXrZtJxUlNTodPpinSw8PLyMqpvQVJSEg4dOoQffvhBb3uTJk2wefNmtGzZEnK5HCtWrEDHjh1x/fp1NGrUyOCxIiMjMXfuXJPiJ4QYj+PY/w9PM36fq3dScP9pVoXF1K9zfXhU0HBFjmNQqrRwdpAU2y9AWMxgfaFAAFspTeFibdHR0dixYwcmTJiAkSNHWjucCmPy/7StW7ciPDwcr732Gl577TV+u1qtxq5duzBixAiTjvdiFs2M7Dy0efNmuLi4YMCAAXrb27dvj/bt2/PPO3bsiODgYKxatQorV640eKwvvvgCERER/HO5XA5/f38TakEIKQlj+de9psxOn1xoIh5jF8YxhlAANKnnVu4EQKXRFTsDolqjg0wqQoPaLmaNnVjG5cuXsWPHDjDGEBAQYO1wKpTJScB7772Hnj17wtPTU297dnY23nvvPaOTAA8PD4hEoiJX/SkpKaUOv2CMYePGjRg+fHip6zcLhUK8/PLLuHv3brFlpFIppFKaR5uQiqLjGGLj05CepTS6Z3ri8xwA+U3jHw1qVek6ZGXnquHkIIGkmPpIJDaUAFRB2dnZmD17NhhjGDhwIDp37mztkCqUyUlAcVfqT5480eu0VxqJRIKQkBAcPXoUAwcO5LcfPXq01CEYp06dwr179zBmzBij4r127Vq1v69DSGV2JyEDJ2OelGmiGi83u0qXAAAABIC7sy283CrntMakbL7++ms8e/YM/v7+mDp1qrXDqXBGJwFBQUH8mPvXX38dNjb/7arT6RAfH4+ePXua9OYREREYPnw42rRpg9DQUKxbtw4JCQkYN24cgPxm+qdPnxaZhXDDhg1o164dWrRoUeSYc+fORfv27dGoUSPI5XKsXLkS165d4yd6IIRYXlJaLvKXrjFdswB3c4djFowxiGgS/mrlyJEjOHz4MIRCIebNmwc7u+qf4BmdBBTce7927RrCwsLg4ODAvyaRSFCvXj2ThwiGh4cjLS0N8+bNQ1JSElq0aIGDBw/yvf2TkpKQkJCgt09WVhb27duHFStWGDxmZmYmxo4di+TkZDg7OyMoKAinT59G27ZtTYqNEGI+ykKr6b0W4g/fWvZG7WcnE8PRzjpL/hqDkoDqIz09HYsWLQIAjBkzpsa0HgsYM23GjC1btiA8PLxaT74jl8vh7OyMrKwsODk5WTscQqq83X/G4dilxxAJBXineyB8azmUvlMlxhhDWpYSTQPc4OxA/YmqA8YYfv31Vxw6dAirVq3Sa+2uLCri3GRyLavzUAlCSMXIU+lQMCzelNnwKoJCqUG2QgNROcJgDJBKRLAR0SQ91YVAIEC/fv3Qt2/fytkHpYKYnATodDosW7YMP/74IxISEqBW60/kkZ6ebrbgCCHVg1KlBf6/V4DMyj3mlWodPFxk8HY37pZEcQQCAexkle9qkZgmNjYWtWvX5q+sa1ICAJQhCZg7dy6+//57REREYNasWfjyyy/x8OFDHDhwAF999VVFxEgIqUQ0Wh2yC62iZ4zMHBU/P54pLQGMMaNWxzOFVsfBxVFGzfgEaWlpmDJlCsRiMaKiolCvXj1rh2RxJicBO3bswPr16/HGG29g7ty5eOedd9CgQQO0atUKFy5cwKRJkyoiTkJIJZGRrcL9J5kwpa9/Znb+Kp5CgQBiE5rQ0+RKs683IJOIIKXx+zUex3GYNWsW0tPT0bBhQ/j4+Fg7JKswOQlITk7me006ODggKyt/Ws8+ffpg1qxZ5o2OEFLpaLQ6CCCAh4spM+4JYCMSQioRmdTcyhiDn5cjXMx81U7T8pItW7bg4sWLkMlkiIyMrLETxpn8TahduzaSkpJQp04dNGzYEEeOHEFwcDAuXbpUYz9EQmoSpUoHkci0q3OVOn+IoClX4DqOQSgUwtleAntbsUnvR0hJrl27hjVr1gDIXya4uk8NXBKTu7YOHDgQx44dAwBMnjwZs2bNQqNGjTBixAiMHj3a7AESQioXpVoLGxOWrmWMQa35/yTAhP4AOh0HG5HA6GmGCTFGeno6pk+fDo7j0LNnT/Tt29faIVmVyS0Bixcv5v8+ePBg+Pv746+//kLDhg3Rr18/swZHCKlctDoO/9xLxZOUXNgY2RrAMQbu/6cjEQmFyFMaXnTnRQqVBjJaUpeY2cqVK5Gamor69etjxowZNW40wItMSgI0Gg3Gjh2LWbNmoX79+gCAdu3aoV27dhUSHCGkcnmakoPjMU8gLMMaAFodB4lYCA7GdfeXSW3gYCemWfmIWX3yySfQaDT44IMPasS0wKUxecZAFxcXXLlyhU8CqiOaMZBUVxnZSjxNyYFp3/r/3H2cgUPnH5ZpkhzGgKE9m6BDS+N7YQuFghp/pUZIgUoxY+DAgQNx4MABREREmCUAQkjJshVqJKbmgtNx5T6WQqWFTsfKvMRtZraKvzLv9nIdBPgav3Jobp4aLRt4QESz7BELS0hIwKVLlzBo0CBKKl9gchLQsGFDzJ8/H+fOnUNISAjs7fVn3aJ5AkhNk5GtRFpmXoUdPydPA6VaB6kZ7o3biIRwti97T3uVRsf/iHq52cHJ3rjFfXQ6DlodR538iMUplUp89tlnuHfvHrKzszFq1Chrh1SpmJwEfP/993BxcUFMTAxiYmL0XhMIBJQEkBpFrdHhybNs5OZpK2weeRuRAB7OpozJrzjy3P+mCXe2N35IsJZjEIkERncmJMQcGGNYtGgR7t27Bzc3N7zxxhvWDqnSMTkJiI+Pr4g4CDELtUZnZLcz80jLzEO2QgMPF1uzz2xXETjGkJWt4nvrmyojWwUg/169RquDNte4WxR5Ki3s7cTU059Y1I4dO3Dw4EEIhUIsXrwYtWrVsnZIlQ5Nm0WqjawcFR4myaEzw71zY2l0DPYycZVJAH788w6ePs8p13F0HIOroxRSqfEndIlEAnsZ9fQnlnP+/HmsXLkSADBt2jQEBwdbOaLKiZIAUm2o1Drk5mksujCMTABIqsjVbWpGXrkTACC/ibWOtyOa1/cwQ1SEmF9CQgK++OILcByH/v3746233rJ2SJUWJQGk2lCqdRAJBbQ4TDESU3P5v9f2dCjzfPwcx9C9bT0zRUWI+d24cQMKhQKtWrXC559/TiMCSkBJAKk2cvPUdM+5BElp/7UCdA6qDR93+xJKF+95Zh7cnGmdEFJ59enTB56enmjQoAEkEuNGsNRUlASQakGj5aDS6CCuZkPQGGO4fi8Vtx6kQastX1+HzJz8Tn02IiE8TVoB8D86jkEkFFCyRSoljUYDsTh/CGzbtm2tHE3VUKYk4MyZM1i7di3u37+PvXv3ws/PD9u2bUNAQAA6depk7hgJKZVGq4Nay8HJzrJ5LccYzlx9imcZigo5vkbLITktt/SCJvBysyvzhD1aHUdD/UildOjQIWzcuBFLlixBnTp1rB1OlWHyL+a+ffswfPhwDB06FFevXoVKlX91kZ2djUWLFuHgwYNmD5KQ0qg0Ouh0rMLG6hfnTkIGLv/7zCLvZY5WDqlYhCZ1XfE8s2xJC8cYnOyk1BJAKpUrV65g3rx50Gg0OHz4MMaOHWvtkKoMk5OABQsW4LvvvsOIESOwa9cufnuHDh0wb948swZHiLHUGg4Ci84QkC8p1bxX6YbYSm3QubUfWjQof2/8rBwV7Gxt4F3G/gAAIBYJaagfqTQSEhIwbdo0aDQavPbaa3j//fetHVKVYnISEBcXh86dOxfZ7uTkhMzMTHPERIjJFEoNhCascW8uKRn/TRc8dkBL2ErMfztCKBKYbR4CnY7BTmoDV0eZWY5HiDVlZmZi8uTJkMvlaNGiBebPn2+V34GqzORfLB8fH9y7dw/16tXT23727NlqvbIgqdwUSg3EFr4VwDGG5//fF8DRTgJHu8rfC1nHGKQVkKgQYmlqtRqffPIJHj9+DF9fXyxduhRSKY1aMZXJv5offvghJk+ejL///hsCgQCJiYnYsWMHpk2bho8//rgiYiSkRBqtDmoNZ/GRAfIcNVQaHQDA07VyzO1fOlbtRlCQmikqKgrXr1+Hg4MDli9fDjc3N2uHVCWZfEnw2WefISsrC127doVSqUTnzp0hlUoxbdo0TJgwoSJiJFaSLlfiWXrF9Ho3K8agUutg52jZK9yUQiMCPF3tLPreZcEYgwACi3eeJKQijBw5Erdu3cLYsWOpFbocBIyVbSURhUKB2NhYcByHZs2awcHBwdyxWY1cLoezszOysrLg5ORk7XCs5v6TTDxLV0BWBZqPRUIBbGU2yMpWoYxr45js6p0UXLv7HADQ/5UGaOjvYpk3LiOtlkN2ngbN67vBTlb25YQJqSwYYzVqNsCKODeZ/Ou+ZcsWDB48GPb29mjTpo1ZgiCVj1bHISdPA3tbMeyrwAmD4xh2HYlDkpnH1BurllvF3Q5gjEGp1qGM+TpPo+NgIxLQ7QBSZR0/fhzZ2dno378/ANSoBKCimJwEFNz779u3L4YNG4aePXvCxqbyXykS0yhVWqjUOjjbV/7ObgCQlJZrtQTA0U4CqY0IKrXO7MdmYMhRqCGRiCAqZ69nG5EQtlIbuh1AqqQrV65g5syZUKvVqFWrFjp06GDtkKoFk8/eSUlJOHz4MHbu3IkhQ4bA1tYWb731FoYNG0b/KNVInkoHHceVeWY5S3v8LJv/ex0vRzhaKHmxEQrh5+UAtc78CUABZwcp6ng7QSYtf7ItFNDVE6l67ty5g6lTp0KtVqNr165o3769tUOqNkz+VbGxsUGfPn3Qp08fKBQK7N+/Hz/88AO6du2K2rVr4/79+xURJ7GwnDw1RIKqkQAA+klA93Z1y7xCXlk8z1TA18MB7s4VM/ZeKKTJeUjNlZiYiIkTJyI3NxdBQUFYuHAhzQVgRuW6tLCzs0NYWBgyMjLw6NEj3L5921xxESvScQzZuflN0FWBVsfxy+Q62UuscgtDKhbRVLqEmFlGRgYmTJiAtLQ0NGzYEEuXLqVVAc2sTOmUQqHAjh070Lt3b/j6+mLZsmUYMGAAbt68ae74iBWo1FqoNByk4qpxUktKy4VWl7/Cnr+no0WbuznGIBQIqsxtE0KqCpVKhQkTJiAhIQE+Pj5YtWoVHB0drR1WtWNyS8A777yDX3/9FXZ2dnjrrbdw8uRJ6gtQzeSptNDqdBDbVI3ZtwrfCvD3suyPhE7HIBIJaVU9QsxMIpGga9eueP78OVavXo1atWpZO6RqyeQkQCAQYPfu3QgLC6NRAdVUbp4WAoEAsfFpSE6r/JMFxSdm8X+3fBLAQSSkYXeEmJtAIMD777+PwYMHw8XFxdrhVFsmn8V/+OGHioiDVBKMMchzVUjLysOh8w+tHY5JXBykcLJwfwAtV9ASQEkAIeWl0WiwYcMGjBgxAnZ2+bNwUgJQsYxKAlauXImxY8dCJpNh5cqVJZadNGmSWQIj1qFU66BS65CaqbR2KCZr1bD8S+2aSqfjYCsV07A7QspJp9Phyy+/xPHjx3Hjxg1ERUXR98oCjEoCli1bhqFDh0Imk2HZsmXFlhMIBNUqCXiWnguFpmp0jjMXjZaDWqtDVo6K39anYwCcLTjkrixkEhu4OFo+Rp2OQVZFRlEQUllxHIe5c+fi+PHjEIvFGDFiBCUAFmJUEhAfH2/w79XdoyQ57LMtNBF9JSIT2yAjOz8JEAgEaFDbhZq7i6FjHC3NS0g5cByHBQsW4ODBgxAKhVi8eDFNBmRBJv96zZs3D9OmTePv1xTIy8vD//73P3z11VdmC87a3J1t4ehY+VeHMzfGGNLl+bcDnO0llACUgj4fQsqG4zjMnz8fv/76K4RCIRYsWIAuXbpYO6waxeRfr7lz5yInJ6fIdoVCgblz55olKGJdOXkaqDX50+C6OVXMLHjVBS3NS0jZLVu2jE8AFi5ciB49elg7pBrH5F+v4pZuvH79Otzc3EwOIDo6GgEBAZDJZAgJCcGZM2eKLXvy5EkIBIIij3///Vev3L59+9CsWTNIpVI0a9YM+/fvNzmumiw9679OgW4VNBVuRVBrdHieqcDzzDyLPYQiAWxoeCAhZdKvXz+4u7sjMjIS3bt3t3Y4NZLRtwNcXV35k27jxo31EgGdToecnByMGzfOpDffvXs3pkyZgujoaHTs2BFr165Fr169EBsbizp16hS7X1xcnN5ayoUnkTh//jzCw8Mxf/58DBw4EPv378fbb7+Ns2fPol27dibFV1MV3AoAqlZLQFauCl7u9hZd+lgAwJY6BhJSJo0aNcKBAwdga1txS3GTkgmYkYuUb9myBYwxjB49GsuXL4ezszP/mkQiQb169RAaGmrSm7dr1w7BwcFYs2YNv61p06YYMGAAIiMji5Q/efIkunbtioyMjGLHjoaHh0Mul+PQoUP8tp49e8LV1RU7d+40uI9KpYJK9V9veLlcDn9/f1y59QiOjk4G96nO/ryUgOt3nwMA3ukeCN9aDlaOqHRKlRYqjQ5NA9xgZ8EkgBBiPK1Wi8jISLzxxhsIDg62djhVjlwuh7OzM7KysvQuhMvD6JaAkSNHAgACAgLQoUMHiMXl+6FVq9WIiYnB9OnT9bb36NED586dK3HfoKAgKJVKNGvWDDNnzkTXrl35186fP4+pU6fqlQ8LC8Py5cuLPV5kZCT1ZyikKrYEZOdp4ONuRwkAIZWUWq3G9OnTcfr0aZw4cQK//PILHBwq/wVGdWdUEiCXy/msIygoCHl5ecjLyzNY1tjsJDU1FTqdDl5eXnrbvby8kJycbHAfHx8frFu3DiEhIVCpVNi2bRtef/11nDx5Ep07dwYAJCcnm3RMAPjiiy8QERHBPy9oCaipCvoESMUiZOdpkJOnsXJEpZOIhajlWvNGchBSFSgUCkRERODy5cuQSCSYN28eJQCVhFFJgKurK5KSkuDp6QkXFxeDHQMLOgzqdDqTAnjxWMV1PASAwMBABAYG8s9DQ0Px+PFjfPvtt3wSYOoxAUAqlUIqrdyT4ViKUq1FrjL/pO9oL4FvLXvYSSv/1bXYRgh728ofJyE1jVwux6RJk3Dz5k3Y2dlh+fLldCugEjEqCTh+/Djf8//EiRNmeWMPDw+IRKIiV+gpKSlFruRL0r59e2zfvp1/7u3tXe5j1mSFbwU4O0hgLxPDw4U67RBCTJeWlobx48fj3r17cHJywurVq9GsWTNrh0UKMSoJKDx5g7kmcpBIJAgJCcHRo0cxcOBAfvvRo0fRv39/o49z9epV+Pj48M9DQ0Nx9OhRvX4BR44coeWOjVR4eKCrg5TGwBNCymzr1q24d+8ePDw8EB0djfr161s7JPICk2cMPHz4MBwcHNCpUycAQFRUFNavX49mzZohKioKrq6uRh8rIiICw4cPR5s2bRAaGop169YhISGBH2r4xRdf4OnTp9i6dSsAYPny5ahXrx6aN28OtVqN7du3Y9++fdi3bx9/zMmTJ6Nz5874+uuv0b9/f/z888/4888/cfbsWVOrWiMVbglwdZLRGHhCSJlNmDABeXl5GDlyJPz8/KwdDjHA5F/4Tz/9FHK5HADwzz//ICIiAr1798aDBw/0OtcZIzw8HMuXL8e8efPQunVrnD59GgcPHkTdunUBAElJSUhISODLq9VqTJs2Da1atcIrr7yCs2fP4vfff8egQYP4Mh06dMCuXbuwadMmtGrVCps3b8bu3btpjgAjFSQBjOVPFGQjokU8CCHGu3PnDjiOAwCIxWLMmDGDEoBKzOh5Ago4ODjg5s2bqFevHubMmYObN29i7969uHLlCnr37l1iL/yqomAsZk2cJ2DjrzeRka2CSCjA+/1boGUDD4jolgAhxAhHjhzB7NmzMWjQIEybNo1WAjQzq84TUEAikUChUAAA/vzzT4wYMQIA4ObmxrcQVBeZ2SpombL0gtWEjjFk5agBAC4OUohthJQAEEJKxRjDjh07+PlY0tLSwHEcRCKaTbOyMzkJ6NSpEyIiItCxY0dcvHgRu3fvBpDfBFS7dm2zB2hN2w/fhkRmb+0wrMLJQQIZLZFLCCkFx3FYtmwZPyPru+++iylTpkAopAuIqsDkf6XVq1fDxsYGe/fuxZo1a/h7PYcOHULPnj3NHiCxDg9nW8hoTnxCSAlUKhW++OILPgGYOnUqIiIiKAGoQkzuE1ATFNx3+XrTGdja1bxZrVwdpajn44jGddzg6Uaz8BFCimKMYdy4cYiJiYGNjQ3mzZtHSwFXsErRJwDIXzXwwIEDuH37NgQCAZo2bYr+/ftXu/s/3dvWqXEdAws8z1RATMMDCSHFEAgEeOedd/DgwQN8/fXXNAtgFWVyEnDv3j307t0bT58+RWBgIBhjuHPnDvz9/fH777+jQYMGFREnsSCOMQgFApojgBBSRG5uLuzt8/tKvfrqq2jbti3s7KjFsKoy+Vd+0qRJaNCgAR4/fowrV67g6tWrSEhIQEBAACZNmlQRMRIL0+kYbERCmi2QEMJjjGHTpk0YNGiQ3lBwSgCqNpNbAk6dOoULFy7wawkAgLu7OxYvXoyOHTuaNThiHTodB5FIQEkAIQRA/kRtCxcuxO+//w4gf3r34cOHWzkqYg4mJwFSqRTZ2dlFtufk5EAikZglKGJdWh0HkUhIswUSQpCcnIzPPvsMsbGxEAqF+OyzzzB48GBrh0XMxORLvT59+mDs2LH4+++/wRgDYwwXLlzAuHHj0K9fv4qIkViYlmOQSkU02xchNdylS5cwbNgwxMbGwtnZGatXr6YEoJoxOQlYuXIlGjRogNDQUMhkMshkMnTs2BENGzbEihUrKiJGYmE6HQeZmCYKIqQm++uvvzB+/HhkZmYiMDAQ27ZtQ9u2ba0dFjEzk3/pXVxc8PPPP+Pu3bu4ffs2AKBZs2Zo2LCh2YMj1sExQCKm/gCE1GQhISFo2LAhGjVqhBkzZkAqlVo7JFIByny516hRI/7ET83G1Q2jOQIIqYGSkpLg5eUFoVAImUyGdevWwd7enn7jq7Ey/dJv2LABLVq04G8HtGjRAt9//725YyNWwBiDQAAaGUBIDfPHH38gPDwcGzdu5Lc5ODhQAlDNmdwSMGvWLCxbtgwTJ05EaGgoAOD8+fOYOnUqHj58iAULFpg9SGI5Oo5BKKQ5AgipKVQqFb799lvs378fABATE4PRo0fT/P81hMlrB3h4eGDVqlV455139Lbv3LkTEydORGpqqlkDtIaC+Zmv3HpU46YNVql1UGm0aN7AA1Jx9ZoGmhCi7+HDh/j8889x//59CAQCjBkzBh988EG1mwK+uqgUawfodDq0adOmyPaQkBBotVqzBEWsR8sVzBFAVwGEVGe///47IiMjoVQq4ebmhgULFlDv/xrI5CRg2LBhWLNmDZYuXaq3fd26dRg6dKjZAqsM1GoOKrXO2mFYlFqtg4OdGCIh3QckpLp69uwZFi5cCLVajbZt22L+/Plwd3e3dljECso0OmDDhg04cuQI2rdvDwC4cOECHj9+jBEjRiAiIoIv92KiUNWodFqIa1jrhshGAHtbmvmRkOrMy8sL06ZNQ3p6Ot3/r+FM7hPQtWtX4w4sEOD48eNlCsraCu67PE9NN9t9l6pEJBJSSwAh1YhGo8H69evRqVMntGrVytrhkDKqiD4BJicBNUFFfNCEEGINDx8+xKxZs3D79m34+flhz549tM5LFVUpOgYSQgip/Bhj2LdvH5YtWwaVSgUnJydMnDiREgCih5IAQgipZhITE7FgwQJcvHgRANC2bVvMmTMHnp6eVo6MVDaUBBBCSDWSkJCAYcOGQaFQQCKRYMKECRgyZAh1/iMGURJACCHViL+/P4KDg5GTk4OvvvoKderUsXZIpBKjJIAQQqowjuOwf/9+dO/eHU5OThAIBFi4cCFsbW3p6p+Uqkz/Q7Zt24aOHTvC19cXjx49AgAsX74cP//8s1mDI4QQUrz79+/j/fffR2RkpN68LPb29pQAEKOY/L9kzZo1iIiIQO/evZGZmQmdLn9GPRcXFyxfvtzc8RFCCHmBUqnEqlWr8O677+LGjRuws7NDixYtQCO+ialMTgJWrVqF9evX48svv9RbZKJNmzb4559/zBocIYQQfX/99RfefvttbNmyBTqdDl27dsXevXsxePBgWvaXmMzkPgHx8fEICgoqsl0qlSI3N9csQRFCCCnqp59+wqJFiwAA3t7e+Oyzz9C5c2crR0WqMpNbAgICAnDt2rUi2w8dOoRmzZqZIyZCCCEGdOvWDbVq1cKwYcPw448/UgJAys3kloBPP/0U48ePh1KpBGMMFy9exM6dOxEZGYnvv/++ImIkhJAa6cKFCzh27BhmzJgBgUAAJycn/PTTT7C1tbV2aKSaMDkJeO+996DVavHZZ59BoVDg3XffhZ+fH1asWIEhQ4ZURIyEEFKjPHnyBEuXLsXp06cBAKGhoXjttdcAgBIAYlblWkAoNTUVHMdVu6koaQEhQog1KBQKbNq0Cdu3b4dGo4FIJEJ4eDg++OADODo6Wjs8YmWVbgEhDw8PswRBCCE1Gcdx+PXXX7FmzRqkpqYCANq3b49PPvkEAQEBVo6OVGcmJwEBAQElDkN58OBBuQIihJCaaMeOHUhNTYWfnx8iIiLQuXNnGvJHKpzJScCUKVP0nms0Gly9ehWHDx/Gp59+aq64CCGkWouNjUWDBg0glUohFArxySef4P79+xg8eDAt90ssplx9AgqLiorC5cuXsWnTJnMczqqoTwAhpKI8ffoUUVFROHLkCCZNmoQRI0ZYOyRSRVTEuclsk0v36tUL+/btM9fhCCGkWklJScHXX3+NN998E0eOHIFAIMCzZ8+sHRap4cy2iuDevXvh5uZmrsMRQki1kJaWhi1btmDv3r1Qq9UAgHbt2mHy5Mlo3LixlaMjNZ3JLQFBQUEIDg7mH0FBQfDx8cGMGTMwY8YMkwOIjo5GQEAAZDIZQkJCcObMmWLL/vTTT+jevTtq1aoFJycnhIaG4o8//tArs3nzZggEgiIPpVJpcmyEEFJey5cvxw8//AC1Wo2goCCsW7cOUVFRlACQSsHkloABAwboPRcKhahVqxZeffVVNGnSxKRj7d69G1OmTEF0dDQ6duyItWvXolevXoiNjUWdOnWKlD99+jS6d++ORYsWwcXFBZs2bULfvn3x999/661n4OTkhLi4OL19ZTKZSbERQkhZyOVyaLVavmX0vffew9OnT/Hhhx+ibdu21OOfVComdQzUarXYsWMHwsLC4O3tXe43b9euHYKDg7FmzRp+W9OmTTFgwABERkYadYzmzZsjPDwcX331FYD8loApU6YgMzOzzHFRx0BCiKlSU1OxY8cO7N27F927d+d/kwgxF6t3DLSxscFHH30ElUpV7jdWq9WIiYlBjx499Lb36NED586dM+oYHMchOzu7SF+EnJwc1K1bF7Vr10afPn1w9erVEo+jUqkgl8v1HoQQYozExERERkaiX79+2LZtG/Ly8nD37l1oNBprh0ZIqUy+HdCuXTtcvXoVdevWLdcbp6amQqfTwcvLS2+7l5cXkpOTjTrGkiVLkJubi7fffpvf1qRJE2zevBktW7aEXC7HihUr0LFjR1y/fh2NGjUyeJzIyEjMnTu37JUhhNQ48fHx2LRpEw4fPgyO4wAArVq1wpgxY9ChQwdq9idVgslJwMcff4xPPvkET548QUhICOzt7fVeb9WqlUnHe/GLwhgz6suzc+dOzJkzBz///LPe2gXt27dH+/bt+ecdO3ZEcHAwVq1ahZUrVxo81hdffIGIiAj+uVwuh7+/v0n1IITULH/++ScOHjwIIP93Z/To0QgKCqKTP6lSjE4CRo8ejeXLlyM8PBwAMGnSJP41gUDAn7x1Op1Rx/Pw8IBIJCpy1Z+SklKkdeBFu3fvxpgxY7Bnzx5069atxLJCoRAvv/wy7t69W2wZqVQKqVRqVNyEkJpHq9Xi2LFjcHd3R5s2bQAA4eHhiI+Px7Bhw9CsWTMrR0hI2RidBGzZsgWLFy9GfHy8Wd5YIpEgJCQER48excCBA/ntR48eRf/+/Yvdb+fOnRg9ejR27tyJN954o9T3YYzh2rVraNmypVniJoTUHNnZ2di/fz927dqFlJQUvPTSS9iwYQOA/FFIixYtsnKEhJSP0UlAwSCC8vYFKCwiIgLDhw9HmzZtEBoainXr1iEhIQHjxo0DkN9M//TpU2zduhVAfgIwYsQIrFixAu3bt+dbEWxtbeHs7AwAmDt3Ltq3b49GjRpBLpdj5cqVuHbtGqKioswWNyGl0el01DGsCktKSsLBgwdx/PhxKJVKCAQCBAYG4tVXX4VCoYBQaLbJVgnhicViiEQii76nSX0CzH2vKzw8HGlpaZg3bx6SkpLQokULHDx4kE80kpKSkJCQwJdfu3YttFotxo8fj/Hjx/PbR44cic2bNwMAMjMzMXbsWCQnJ8PZ2RlBQUE4ffo02rZta9bYCTGEMYbk5ORyDVEl1iWXy5Gbm4umTZuiadOmsLGxgb29PWxtbSEQCPDo0SNrh0iqMRcXF3h7e1usb4nR8wQIhUI4OzuXGlh6erpZArMmmieAlFVSUhIyMzPh6ekJOzs76iRWBWi1WggEAv4KLDs7G8+ePYOdnR1cXFz4kz8hFYkxBoVCgZSUFLi4uMDHx6dImYo4N5nUEjB37ly+2Z0Qok+n0/EJgLu7u7XDISUo+MHNyMhAdnY2atWqBQ8PDwD5/ZWcnZ1pOV9icba2tgDyO8h7enpa5NaASUnAkCFD9IbjEUL+U9AHwM7OzsqRkOJotVpkZmYiKytLb9KzwmuLCIVCSgCI1RT8fmg0msqVBFBzGCHGoe9K5cMYw9OnT5Gdnc13ci64xenq6kpri5BKw9K/HyaPDiCEkKpArVbzV/QFq4kyxvh7/Y6OjhbviU1IZWN0ElAwLSYhhFRWWq0WWVlZkMvlyMvLQ/369fmrfA8PD3h4eNDEYIQUYvK0wYQQUpnodDpkZ2cjKysLCoWCb7UUCATIy8vjkwA6+RNSFM14QQixuldffRVTpkwxeT+lUok7d+4gMTERubm5YIzB1tYW3t7eaNSoEVxdXY06TlpaGjw9PfHw4UOTYyDEWIMHD8bSpUutHYYeSgIIIQCAUaNG8ffObWxsUKdOHXz00UfIyMiwdmgA/ls6PCsri98mlUohFAohlUpRq1YtNGzYEAEBAXBzc4ONjfENnZGRkejbty/q1aunt/3cuXMQiUTo2bOnwf2KS14OHDhQpINXcnIyJk6ciPr160MqlcLf3x99+/bFsWPHjI6zLKKjoxEQEACZTIaQkBCcOXPGLPuYq4y5Veb6fvXVV1i4cGHlWq6ekSKysrIYAJaVlWXtUEgVkpeXx2JjY1leXp61QymTkSNHsp49e7KkpCT2+PFj9scffzA/Pz82ZMiQCn/vLl26sMmTJxfZrtPpmFwuZ0+ePGG3b99mt27dYnfu3GEcx/Fl1Gq13nNTKRQK5uLiws6dO1fktTFjxrDJkycze3t79ujRI6Pj3r9/Pyv88xofH898fX1Zs2bN2J49e1hcXBy7efMmW7JkCQsMDCxz7KXZtWsXE4vFbP369Sw2NrbEupiyj7nKlKZLly5s06ZN1aq+wcHBLDo6uth4SvodqYhzEyUBBlASQMqipC+vQqEo9qFSqYwuq1QqjSpbFiNHjmT9+/fX2xYREcHc3Nz454cOHWIdO3Zkzs7OzM3Njb3xxhvs3r17evt06dKFTZw4kX366afM1dWVeXl5sdmzZ/Ov5+TksOHDhzN7e3vm7e3Nvv322yIn09TUVPbee+8xNzc3JpFIWFBQENu1axeLi4tjSUlJTKfTsS5durAJEyawyZMnMxcXF+bp6cnWrl3LcnJy2KhRo5iDgwOrX78+O3jwYIn13rdvH/Pw8CiyPScnhzk6OrJ///2XhYeHs7lz5xYpY2wS0KtXL+bn58dycnKKlM3IyCgxvvJo27YtGzdunN62Jk2asOnTp5drH3OVKY2pSUBVqO+cOXPYK6+8Umw8lk4C6HYAIRbwyiuvFPv49NNP9cp279692LITJ07UK9u3b1+D5czhwYMHOHz4MMRiMb8tNzcXERERuHTpEo4dOwahUIiBAwcWGT20ZcsW2Nvb4++//8Y333yDefPm4ejRowCATz/9FCdOnMD+/ftx5MgRnDx5EjExMXrDkD///HP8/vvvWLRoEQ4cOIBGjRrho48+gru7O7y9vfkFfLZs2QIPDw9cvHgREydOxEcffYS33noLHTp0wJUrVxAWFobhw4dDoVAUW8/Tp0/zywMXtnv3bgQGBiIwMBDDhg3Dpk2byjRUOj09HYcPH8b48eNhb29f5HUXF5di9120aBEcHBxKfBTX3K1WqxETE4MePXrobe/RowfOnTtX5n3MVcbcqkp927Zti4sXL+pNVmVNNDqAEML77bff4ODgAJ1Ox8+iV7gj05tvvqlXfsOGDfD09ERsbCxatGjBb2/VqhVmz54NAGjUqBFWr16NY8eOITQ0FBs2bMCWLVvQqVMnZGdnY+7cuejYsSO0Wi2A/ERj69atWLFiBUaMGAGZTIZu3bqhXr162Lhxo17S9NJLL2HmzJkA8lcdXbx4MTw8PPDBBx8AyL8Hu2bNGty4cQPt27c3WOeHDx/C19e3yPYNGzZg2LBhAICePXsiJycHx44dQ7du3Uz6TO/duwfGGJo0aWLSfgAwbtw4vP322yWW8fPzM7g9NTUVOp0OXl5eetu9vLz4FVjLso+5yhiyaNEiveWZ8/LycOHCBUyYMIHfdujQIYOJblWpr5+fH1QqFZKTk826Km9ZURJAiAWU1DnpxQlrCq6YDXlxCdtff/21fIG9oGvXrlizZg0UCgW+//573LlzR6/14f79+5g1axYuXLiA1NRUvgUgISGhSBJQmI+PD549e4YbN25ArVbD29sb8fHxAACZTIZ69epBp9Px76HRaNC7d29+LnWxWIy2bdvi9u3besct/D4ikQju7u5o2bIlv63gRzklJaXYOhceRlggLi4OFy9exE8//QQAsLGxQXh4ODZu3GhyEsAKDVk0lZubG9zc3Ezer7AX35cxVmosxuxjrjKFvZj0DB06FG+++SYGDRrEbysu6Snrexq7j7nKFPyfLql1ypIoCSDEAgq++NYsawx7e3s0bNgQALBy5Up07doVc+fOxfz58wHk337w9/fH+vXr4evrC47j0KJFC6jVar3jiMViMMbAcRxEIhEEAgG0Wi2SkpIA5E/qIxQK+SZtqVTKn4iLO2ka+kEtfKuiYJ/C2wrKlzTZmYeHR5EREBs2bIBWq9U74TDGIBaLkZGRwQ89dHJy0hutUCAzM5Nf5a1Ro0YQCAS4ffs2BgwYUGwchrx4ZWxIcVfGHh4eEIlERa5EU1JSilyxmrKPucoY8mLSY2trC09PT/7/ZEmqSn0LVtqtVatWqXWyBOoTQAgp1uzZs/Htt98iMTERaWlpuH37NmbOnInXX38dTZs2LXLy1Gg00Gq1yMnJwZ07d/R+FEUiEZo0aQKxWIzExEQ0btwYtWvXBmMMd+/e5cs1bNgQEokEZ8+e1Tvu5cuX0bRpU7PXMSgoCLGxsfxzrVaLrVu3YsmSJbh27Rr/uH79OurWrYsdO3bwZZs0aYLLly8XOealS5cQGBgIIP/EFhYWhqioKOTm5hYpm5mZWWxs48aN04vB0MNQfwYgfzXEkJCQIi1LR48eRYcOHcq8j7nKmFtVqe/NmzdRu3ZtftVKqzNbF8NqhEYHkLKoDkMEXxwdwBhjISEhbPz48Uyn0zF3d3c2bNgwdvfuXXbs2DH28ssvMwBs48aN7N69e+zWrVvs5ZdfZsOGDWO3bt1id+/eZYwx1r9/fzZy5EjGGGPjxo1jderUYX/++Sf7559/WL9+/ZiDg4NeL/vJkyczX19fdujQIXbr1i02cuRI5urqytLT0/kyhnrm161bly1btkxvGwC2f//+Yut948YNZmNjwx97//79TCKRsMzMzCJlZ8yYwVq3bs0/j4+PZ7a2tuzjjz9m165dY3FxcWz16tVMKpWyH3/8kS/34MED5u3tzZo1a8b27t3L7ty5w2JjY9mKFStYkyZNio2tvAqGrW3YsIHFxsayKVOmMHt7e/bw4UO+zKpVq9hrr71m0j7mKvOi7OxslpSUVOLjxdE0Va2+I0eOZKNHjy62DjREsBKgJICURXVNAnbs2MEkEglLSEhghw8fZk2aNGFSqZS1atWKnTx5kgFgK1euZLdu3WK3bt1i7dq1Yx988AHLzc3lx+8XTgKys7PZsGHDmJ2dHfPy8mLffPNNkRN6Xl4emzhxIvPw8GBSqZR17NiRXbx4US8ucyUBjDHWvn179t133zHGGOvTpw/r3bu3wXIxMTEMAIuJieG3Xb58mYWFhTFPT0/m5OTE2rRpw3bu3Flk38TERDZ+/HhWt25dJpFImJ+fH+vXrx87ceJEibGVV1RUFP+ewcHB7NSpU3qvz549m9WtW9ekfcxZ5sVYAJT4KO3zqsz1zcvLY05OTuz8+fPFxm/pJEDAGC0P+CK5XA5nZ2dkZWXx9/UIKY1SqUR8fDw/Y1h1oNPpoFAooFAokJubC6VSCYFAgMDAQL6TYlpaGtRqNezt7WFvb18lV+Y7ePAgpk2bhps3bxbpfEmIuURFReHnn3/GkSNHii1T0u9IRZybqGMgIaSIjIwMpKenGxzLLBaLodFo+AV53N3dLR2e2fXu3Rt3797F06dP4e/vb+1wSDUlFouxatUqa4ehh5IAQmoojuOgVCqhUCiQl5cHLy8vSCQS/rWCBEAikcDe3h52dnaws7Mr0iO/upg8ebK1QyDV3NixY60dQhGUBBBSQ2i1WuTm5iIvLw95eXlQKpV6M+A5OjrySUDB321tbU1aiIcQUrXQt5uQaoYxBo1GA6VSCYlEwt9XzMvLw9OnT/XK2tjYwM7ODra2tnpzDkgkEj4hIIRUX5QEEFKFFT7hFzzy8vL42fc8PDz4JMDW1hZ2dnaQyWT8SV8sFpdpJjtCSPVASQAhVUTBfXqBQMCf2NVqNe7fv1+krEAggFQq1WvKt7GxQb169SwVLiGkCqAkgJBKpuDqXq1WQ6lUQqVSQalUQq1WgzEGJycn1K5dG0B+s71IJIJYLIZMJuOv8qVSKQ11I4SUipIAQqyk4GRfcHXv4ODAv3b//n2Dy9aKRCK9k7tAIEDjxo2pSZ8QUiaUBBBiITk5OfyVvVqthkql4he2kclkfBJQ0JTPGOMX1pHJZHzzfmkrlxFCiLEoCSDEDDiOQ2pqKlQqld6CMN7e3vzfk5OTi6y2JxAI9HrwFwgICKCTOyGkwlESQIiRdDod0tPT9ZYA/d///oeYmBg8fvwYLi4u+OSTT2BjYwOhUAiRSAQvLy/+ZO7g4ACtVgupVMo/JBKJwZM9JQCEEEugJICQF9y9exfx8fF48uQJEhMTkZiYiCdPniA5ORkODg44fvw4X/bx48e4d+8egPz79TY2Nvzsei+Osy/cKkAsJy0tDU2bNsXFixdpdASpMIMHD0aHDh0QERFh7VBMQt2HSY2h0+nw7Nkz3LhxA0eOHMHWrVvxv//9D19++aVeuZUrV2LGjBmIjo7GgQMHcPHiRSQmJoLjOOTl5UGhUPBlR40ahZUrV+LAgQPYuXMnatWqBR8fH3h5ecHV1bVKXdGPGjUKAoEAAoEAYrEYXl5e6N69OzZu3Mj3XTCl3ItlbWxsUKdOHXz00UfIyMjQK5eSkoIPP/wQderUgVQqhbe3N8LCwnD+/Hm9cufOnYNIJELPnj2NrldkZCT69u1rMAEo6XivvvoqpkyZUmT7gQMHivy7JicnY+LEiahfvz6kUin8/f3Rt29fHDt2zOg4yyo6OppfbCYkJARnzpwptqxWq8XMmTMREBAAW1tb1K9fH/PmzdP7d4uMjMTLL78MR0dHeHp6YsCAAYiLi9M7jjFlKoIpdTV2P2PrUtp7f/XVV1i4cCHkcnn5KmlpZluPsBqhpYSrHo7jWFpaGrt16xY7fvw427Nnj97rs2bNYi+//DILCQkx+MjJyeHLRkVFsffff5999dVXbO3atey3335jV65cYc+ePWM6na7YGKrDUsI9e/ZkSUlJ7MmTJywmJoYtXLiQOTg4sF69ejGNRmNSuRfLPn78mP3xxx/Mz8+PDRkyRO+9O3XqxNq1a8eOHz/OHj58yP7++2+2aNEi9ttvv+mVGzNmDJs8eTKzt7dnjx49KrVOCoWCubi4sHPnzhl8vaTjGVqqmDHG9u/fzwr/dMbHxzNfX1/WrFkztmfPHhYXF8du3rzJlixZwgIDA0uNsTwK1rBfv349i42NLfWzWbBgAXN3d2e//fYbi4+PZ3v27GEODg5s+fLlfJmwsDC2adMmdvPmTXbt2jX2xhtvsDp16uh9R4wpU5ouXbqwTZs2VVhdjd3PmLoY+97BwcEsOjra6DoZYumlhCkJMICSgMpDp9OxtLQ0FhcXx/766y9+fXrGGNu0aRN77733WN++fVloaGiRE3tubi5fdsGCBSwkJIS1bduW9enTh40ZM4bNmDGDrVixgu3Zs0evbFlVhySgf//+RbYfO3aMAWDr1683qVxxZSMiIpibmxv/PCMjgwFgJ0+eLDG+nJwc5ujoyP79918WHh7O5s6dW2qd9u3bxzw8PMp0PGOTgF69ejE/Pz+DJ8CMjIxSYyyPtm3bsnHjxulta9KkCZs+fbrB8m+88QYbPXq03rZBgwaxYcOGFfseKSkpDAA7depUucq8yNQkwNS6lnU/Q3Ux9hhz5sxhr7zySql1KYmlkwDqE0AsjjGGvLw8ZGRkICMjA82bN+ebV/fs2YPz588jNTWVfxRuqjx16hTs7e0BAE+fPsWNGzf41wQCAdzd3eHl5QVvb28olUrY2dkByF+9a+zYsXB3d7f4JDpLfohBdq669IJm5mgvwSfvhpT7OK+99hpeeukl/PTTT3j//ffLVe7Bgwc4fPiw3kqEDg4OcHBwwIEDB9C+fXt+ieIX7d69G4GBgQgMDMSwYcMwceJEzJo1q8RbLqdPn0abNm3MdrwXpaen4/Dhw1i4cCH//7IwFxeXYvddtGgRFi1aVOLxDx06hFdeecXga2q1GjExMZg+fbre9h49euDcuXMG9+nUqRO+++473LlzB40bN8b169dx9uxZLF++vNgYsrKyAABubm7lKlMeZalrWfd7sS6mHKNt27aIjIyESqUq9v9xZUNJADGLwif1jIwMdOzYkf8x/fHHH3H27Fn+tfT0dL2hcidPnuTHyN+9exenT5/WO7ZAIICrqys8PDyQm5vL/9gOGDAAoaGh8PDwgIeHB2rVqlXsMreFe/RbWnauGlk5Kqu9vzk0adJEL+Eypdxvv/0GBwcH6HQ6KJVKAMDSpUv5121sbLB582Z88MEH+O677xAcHIwuXbpgyJAhaNWqFV9uw4YNGDZsGACgZ8+eyMnJwbFjx9CtW7di43n48CF8fX0NvlaW473o3r17YIyhSZMmRu9TYNy4cXj77bdLLOPn51fsa6mpqdDpdPDy8tLb7uXlheTkZIP7fP7558jKykKTJk0gEomg0+mwcOFCvPPOOwbLM8YQERGBTp06oUWLFmUuAxRNevLy8nDhwgVMmDCB31Zc0lOWupZlP0N1MeUYfn5+UKlUSE5ORt26dYuNqzKhJIDo0Wq1kMvlyMrK0vtTLpdjyJAh/FX0hg0bcOLECWRmZiIjI4Nfe77A8ePH4eTkBCD/h9JQ1i2VSuHm5obc3Fw+CQgLC0OTJk34E7uHhwfc3NwMLmfbvHlzNG/e3Nwfgdk52ltnNT5zvi9jzKgrZEPlunbtijVr1kChUOD777/HnTt3MHHiRL0yb775Jt544w2cOXMG58+fx+HDh/HNN9/g+++/x6hRoxAXF4eLFy/ip59+ApCfOISHh2Pjxo0lnrTz8vKKzMEAoMzHM1RfoGxDOt3c3Mxy5fzie5f0b7V7925s374dP/zwA5o3b45r165hypQp8PX1xciRI4uUnzBhAm7cuIGzZ88W+/7GlAGKJj1Dhw7Fm2++iUGDBvHbSkp6ANPqWpb9SqqLMccoWImzcOfhyo6SgGqG4zi95u779+8jMTEROTk5yMnJQXZ2Nv/3nJwcLFiwgC8/e/Zs/P7778Ueu0+fPvyJPSUlBf/++6/e6xKJBK6urnBzc4NSqeTL9urVCy1atOBfc3V1haurq97StQVCQkIQElL+JuzKxBxN8tZ2+/ZtBAQElKmcvb09GjZsCCB/5EXXrl0xd+5czJ8/X6+cTCZD9+7d0b17d3z11Vd4//33MXv2bIwaNQobNmyAVqvVO0kwxiAWi5GRkQFXV1eD8Xh4eBQZiQDAqOM5OTnxTcOFZWZm8v+3GzVqBIFAgNu3b2PAgAGlfj6Flfd2gIeHB0QiUZGr0ZSUlCJXrQU+/fRTTJ8+HUOGDAEAtGzZEo8ePUJkZGSRJGDixIn45ZdfcPr0aX6tihcZU6bAi0mPra0tPD09+f8bJSlLXU3dr7i6mHKM9PR0ANZteTQVJQGVRMHws9zcXL0/FQoFlEolunfvzpfds2cPbty4YfDErlAocOHCBb0r9iNHjhT7vtOnT+d/0AqumAQCARwdHeHk5AQnJyc4OzvD2dlZ7978m2++ic6dO8PJyYk/sdva2hrMroOCghAUFGSWz4lY3vHjx/HPP/9g6tSpZik3e/Zs9OrVCx999FGxTfUA0KxZMxw4cABarRZbt27FkiVL0KNHD70yb775Jnbs2KHXpFxYUFAQtm/frrfN2OM1adIEhw4dKnLMS5cuITAwEED+iS0sLAxRUVGYNGlSkX4BmZmZxfYLKO/tAIlEgpCQEBw9ehQDBw7ktx89ehT9+/c3uI9CoSjSJ0YkEul9txljmDhxIvbv34+TJ08aTP6MKWNOZamrsfuVVhdT3vvmzZuoXbs2PDw8ylVfS6IkwEQcx+mt3V74odVq0bZtW77soUOHEB8fX+TkrlAowHEcNm7cyJedMmVKiR1cXn/9df7Le/Xq1RJP7Dk5OfyJvW7dumjWrBnf+arww9HRUa+Z/eOPP8bHH38MR0fHUjvPNW7cGI0bNy75wyJVTsH9zII5FQ4fPozIyEj06dMHI0aMMLmcIa+++iqaN2+ORYsWYfXq1UhLS8Nbb72F0aNHo1WrVnB0dMTly5fxzTffoH///vjtt9+QkZGBMWPGwNnZWe9YgwcPxoYNG4pNAsLCwvDFF1/otRYYe7yPP/4Yq1evxvjx4zF27FjY2tri6NGj2LBhA7Zt28bvEx0djQ4dOqBt27aYN28eWrVqBa1Wi6NHj2LNmjW4ffu2wdjMcTsgIiICw4cPR5s2bRAaGop169YhISEB48aNAwCsXr0a+/fv5+cr6Nu3LxYuXIg6deqgefPmuHr1KpYuXYrRo0fzxxw/fjx++OEH/Pzzz3B0dOSvgJ2dnfnWO2PKvKjgQqXArl27AEDvCtvNza3IJFvG1tVQfY3Zz5i6GPPeAHDmzJkiiWWlZ7ZxBtVIwTCMwYMHs/fee0/vtdGjRxc71vzFoSETJkwotmxISIjemPPPP/+chYSEsJdffpm9+uqrrFevXmzw4MFs+PDhbNy4cUypVPJlT506xbZt28b279/P/vzzT3bhwgV28+ZN9ujRI5aWllbiWHZScarDEEEADACzsbFhtWrVYt26dWMbN27U+z9lbLmCsoaGE+7YsYNJJBKWkJDAlEolmz59OgsODmbOzs7Mzs6OBQYGspkzZzKFQsH69OnDevfubTDmmJgYBoDFxMQUW6/27duz7777jn9uyvEuX77MwsLCmKenJ3NycmJt2rRhO3fuLLJfYmIiGz9+PKtbty6TSCTMz8+P9evXj504caLYuMwlKiqKf9/g4GC9oW2zZ89mdevW5Z/L5XI2efJkVqdOHSaTyVj9+vXZl19+yVQqFV+m4N/2xUfh4XzGlHnR7Nmzi92v4FHa51VSXQ3V15j9jK1Lae+dl5fHnJyc2Pnz50usQ2ksPURQwJiB9UprOLlcDmdnZ7Ru3RrOzs44efIk/9r48ePx999/AwC/ulvBw87ODlu3buXL/vjjj3j48CHs7OxgZ2cHe3t72Nra8n+2b9+ev+JWKBQQiUTFziVPKj+lUon4+Hh+VjFSORw8eBDTpk3DzZs3LT48lNQcUVFR+Pnnn0tspTVGSb8jBeemrKwsvrW33MyWTpRRVFQUq1evHpNKpSw4OJidPn26xPInT55kwcHBTCqVsoCAALZmzZoiZfbu3cuaNm3KJBIJa9q0Kfvpp59Miqkg2zp27Bi7fv263mu5ublMqVTqTVpDCGNVvyWgOlu+fDlLSEiwdhikGlu7di37999/y30cS7cEWDUt3r17N6ZMmYIvv/wSV69exSuvvIJevXohISHBYPn4+Hj07t0br7zyCq5evYoZM2Zg0qRJ2LdvH1/m/PnzCA8Px/Dhw3H9+nUMHz4cb7/9Nn/1boo2bdrojVMGADs7O0ilUrpaJ6QKmTx5Mvz9/a0dBqnGxo4dy3cYrUqsejugXbt2CA4Oxpo1a/htTZs2xYABAxAZGVmk/Oeff45ffvlFr6PNuHHjcP36dX6hkfDwcMjlcr1evT179oSrqyt27txpMA6VSqU3zl0ul8Pf39+8TS6k2qPbAYSQ8rL07QCrtQQUTMX4Yk/KkqZzPH/+fJHyYWFhuHz5MjQaTYllSup5HxkZyQ+Dc3Z2pisGQgghNYLVkoCyTAOZnJxssLxWq0VqamqJZUqaWvKLL75AVlYW/3j8+HFZqkQIgP9mkSOEEFNZ+vfD6vMEmDoNpKHyL2439ZhSqbTKLPZAKq+CdQsUCkWxY6UJIaQkBVMOF7cOirlZLQkoyzSQ3t7eBsvb2NjA3d29xDIlTS1JiDmIRCK4uLggJSUFQH4nUupASggxBmMMCoUCKSkpcHFxgUgkssj7Wi0JKMs0kKGhofj111/1th05cgRt2rThs6bQ0FAcPXpUb+rSI0eOoEOHDhVQC0L0eXt7AwCfCBBCiClcXFz43xFLsOrtgNKmYvziiy/w9OlTfgKecePGYfXq1YiIiMAHH3yA8+fPY8OGDXq9/idPnozOnTvj66+/Rv/+/fHzzz/jzz//LHWFK0LMQSAQwMfHB56ennxnVUIIMYZYLLZYC0ABqyYB4eHhSEtLw7x585CUlIQWLVrg4MGD/DrMSUlJenMGBAQE4ODBg5g6dSqioqLg6+uLlStX4s033+TLdOjQAbt27cLMmTMxa9YsNGjQALt370a7du0sXj9Sc4lEIot/mQkhxFQ0bbABFTI1IyGEEFIO1WqeAEIIIYRYFyUBhBBCSA1l9XkCKqOCOyRyudzKkRBCCCH5Cs5J5ryLT0mAAdnZ2QBA0wcTQgipdLKzs+Hs7GyWY1HHQAM4jkPjxo0RExNTJSZ7KVjw6PHjxxbvyPjyyy/j0qVLVeL45vicyhpPWfYzdh9r/vtbm6l1r+j/r+ZUXKyW+PeuTN/r0uprjliryveaMYbs7Gz4+vpCKDTP3XxqCTBAKBRCIpGYLdOyFCcnJ4ufBEQiUYW+Z0UcvzyfU1njKct+pu5jjX//ysLYulf0/1dzKi3Wivz3rozf6+Lqa45Yq9L32tznJeoYWIzx48dbO4QqoaI/p8r271DWeMqyX2Wre3VQlT5Ta8Zalb7X5jhWTf5e0+2AaoDmNTBOdf2cqmu9jFET617T6lzT6lvAUvWmloBqQCqVYvbs2bQSYimq6+dUXetljJpY95pW55pW3wKWqje1BBBCCCE1FLUEEEIIITUUJQGEEEJIDUVJACGEEFJDURJACCGE1FCUBFRC0dHRCAgIgEwmQ0hICM6cOVNi+VOnTiEkJAQymQz169fHd999p/f6Tz/9hDZt2sDFxQX29vZo3bo1tm3bVpFVsAhzf04AkJmZifHjx8PHxwcymQxNmzbFwYMHK6oKBplSr6SkJLz77rsIDAyEUCjElClTDJbbt28fmjVrBqlUimbNmmH//v0VFH35VETdly9fjsDAQNja2sLf3x9Tp06FUqmsoBqUjSn1Pnv2LDp27Ah3d3fY2tqiSZMmWLZsmV6Z9evX45VXXoGrqytcXV3RrVs3XLx4saKrYRRT6nry5EkIBIIij3///VevXGX43pbk9OnT6Nu3L3x9fSEQCHDgwIFS9zHm98os32tGKpVdu3YxsVjM1q9fz2JjY9nkyZOZvb09e/TokcHyDx48YHZ2dmzy5MksNjaWrV+/nonFYrZ3716+zIkTJ9hPP/3EYmNj2b1799jy5cuZSCRihw8ftlS1zK4iPieVSsXatGnDevfuzc6ePcsePnzIzpw5w65du2apaplcr/j4eDZp0iS2ZcsW1rp1azZ58uQiZc6dO8dEIhFbtGgRu337Nlu0aBGzsbFhFy5cqODamKYi6r59+3YmlUrZjh07WHx8PPvjjz+Yj48PmzJlSgXXxnim1vvKlSvshx9+YDdv3mTx8fFs27ZtzM7Ojq1du5Yv8+6777KoqCh29epVdvv2bfbee+8xZ2dn9uTJE0tVyyBT63rixAkGgMXFxbGkpCT+odVq+TKV4XtbmoMHD7Ivv/yS7du3jwFg+/fvL7G8Mb9X5vpeUxJQybRt25aNGzdOb1uTJk3Y9OnTDZb/7LPPWJMmTfS2ffjhh6x9+/Ylvk9QUBCbOXNm+YK1oor4nNasWcPq16/P1Gq1+QM2kqn1KqxLly4GT4Rvv/0269mzp962sLAwNmTIkHLFam4VUffx48ez1157TW9bREQE69SpU7liNafy1LvAwIED2bBhw4p9XavVMkdHR7Zly5Yyx2kOpta1IAnIyMgo9piV4XtrCmOSAGN+r8z1vabbAZWIWq1GTEwMevToobe9R48eOHfunMF9zp8/X6R8WFgYLl++DI1GU6Q8YwzHjh1DXFwcOnfubL7gLaiiPqdffvkFoaGhGD9+PLy8vNCiRQssWrQIOp2uYirygrLUyxjF1b08xzS3iqp7p06dEBMTwzeFP3jwAAcPHsQbb7xRrnjNxRz1vnr1Ks6dO4cuXboUW0ahUECj0cDNza1c8ZZHeeoaFBQEHx8fvP766zhx4oTea9b+3lYEY36vzPW9pgWEKpHU1FTodDp4eXnpbffy8kJycrLBfZKTkw2W12q1SE1NhY+PDwAgKysLfn5+UKlUEIlEiI6ORvfu3SumIhWsoj6nBw8e4Pjx4xg6dCgOHjyIu3fvYvz48dBqtfjqq68qrD4FylIvYxRX9/Ic09wqqu5DhgzB8+fP0alTJzDGoNVq8dFHH2H69OnlDdksylPv2rVr4/nz59BqtZgzZw7ef//9YstOnz4dfn5+6Natm1niLouy1NXHxwfr1q1DSEgIVCoVtm3bhtdffx0nT57kL2Ks/b2tCMb8Xpnre01JQCX04vLFjLESlzQ2VP7F7Y6Ojrh27RpycnJw7NgxREREoH79+nj11VfNF7iFmftz4jgOnp6eWLduHUQiEUJCQpCYmIj//e9/Fv0xMbVe1jpmRTB3nCdPnsTChQsRHR2Ndu3a4d69e5g8eTJ8fHwwa9as8oZrNmWp95kzZ5CTk4MLFy5g+vTpaNiwId55550i5b755hvs3LkTJ0+ehEwmM2vcZWFKXQMDAxEYGMg/Dw0NxePHj/Htt9/ySUBl+d6amzG/6+b4vlASUIl4eHhAJBIVyeRSUlKKZHwFvL29DZa3sbGBu7s7v00oFKJhw4YAgNatW+P27duIjIyskklARX1OPj4+EIvFEIlEfJmmTZsiOTkZarUaEonEzDXRV5Z6GaO4upfnmOZWUXWfNWsWhg8fzl8lt2zZErm5uRg7diy+/PJLs63JXlblqXdAQACA/Do9e/YMc+bMKZIEfPvtt1i0aBH+/PNPtGrVyrzBm8hc/8bt27fH9u3b+efW/t5WBGN+r8z1vaY+AZWIRCJBSEgIjh49qrf96NGj6NChg8F9QkNDi5Q/cuQI2rRpA7FYXOx7McagUqnKH7QVVNTn1LFjR9y7dw8cx/Fl7ty5Ax8fH4v8kJSlXsYoru7lOaa5VVTdFQpFkRO9SCQCy+8UXebjmou56m3o+/y///0P8+fPx+HDh9GmTRuzxFse5qrr1atX+ducgPW/txXBmN8rs32vTepGSCpcwRCaDRs2sNjYWDZlyhRmb2/PHj58yBhjbPr06Wz48OF8+YKhJFOnTmWxsbFsw4YNRYaSLFq0iB05coTdv3+f3b59my1ZsoTZ2Niw9evXW7x+5lIRn1NCQgJzcHBgEyZMYHFxcey3335jnp6ebMGCBZW2XowxdvXqVXb16lUWEhLC3n33XXb16lV269Yt/vW//vqLiUQitnjxYnb79m22ePHiSj1E0Jx1nz17NnN0dGQ7d+5kDx48YEeOHGENGjRgb7/9tkXrVhJT67169Wr2yy+/sDt37rA7d+6wjRs3MicnJ/bll1/yZb7++msmkUjY3r179YbWZWdnW7x+hZla12XLlrH9+/ezO3fusJs3b7Lp06czAGzfvn18mcrwvS1NdnY2/38VAFu6dCm7evUqPzSyLL9X5vpeUxJQCUVFRbG6desyiUTCgoOD2alTp/jXRo4cybp06aJX/uTJkywoKIhJJBJWr149tmbNGr3Xv/zyS9awYUMmk8mYq6srCw0NZbt27bJEVSqUuT8nxvLH3rZr145JpVJWv359tnDhQr0xyZZgar0AFHnUrVtXr8yePXtYYGAgE4vFrEmTJno/opWJueuu0WjYnDlzWIMGDZhMJmP+/v7s448/LnHImTWYUu+VK1ey5s2bMzs7O+bk5MSCgoJYdHQ00+l0fJm6desa/Gxmz55twVoZZkpdv/76a/7fztXVlXXq1In9/vvvRY5ZGb63JSkY6vjiY+TIkYyxsv9emeN7TUsJE0IIITUU9QkghBBCaihKAgghhJAaipIAQgghpIaiJIAQQgipoSgJIIQQQmooSgIIIYSQGoqSAEIIIaSGoiSAEEIIqaEoCSCkCti8eTNcXFysHUa5CAQCHDhwoMQyo0aNwoABAywSDyGEkgBCLGbUqFEQCARFHvfu3bN2aBaRlJSEXr16AQAePnwIgUCAa9eu6ZVZsWIFNm/ebPngjHDy5EkIBAJkZmZaOxRCzIaWEibEgnr27IlNmzbpbatVq5aVorEsb2/vUss4OztbIBJ9VXW5WULMgVoCCLEgqVQKb29vvYdIJMLSpUvRsmVL2Nvbw9/fHx9//DFycnKKPc7169fRtWtXODo6wsnJCSEhIbh8+TL/+rlz59C5c2fY2trC398fkyZNQm5ubrHHmzNnDlq3bo21a9fC398fdnZ2eOutt/SuejmOw7x581C7dm1IpVK0bt0ahw8f5l9Xq9WYMGECfHx8IJPJUK9ePURGRvKvF74dEBAQAAAICgqCQCDAq6++CkD/dsDatWvh5+ent0QsAPTr1w8jR47kn//6668ICQmBTCZD/fr1MXfuXGi12mLrWvAekZGR8PX1RePGjQEA27dvR5s2beDo6Ahvb2+8++67SElJAZDfctG1a1cAgKurKwQCAUaNGgUgfxnfb775BvXr14etrS1eeukl7N27t9j3J6QyoSSAkEpAKBRi5cqVuHnzJrZs2YLjx4/js88+K7b80KFDUbt2bVy6dAkxMTGYPn06v874P//8g7CwMAwaNAg3btzA7t27cfbsWUyYMKHEGO7du4cff/wRv/76Kw4fPoxr165h/Pjx/OsrVqzAkiVL8O233+LGjRsICwtDv379cPfuXQDAypUr8csvv+DHH39EXFwctm/fjnr16hl8r4sXLwIA/vzzTyQlJeGnn34qUuatt95CamoqTpw4wW/LyMjAH3/8gaFDhwIA/vjjDwwbNgyTJk1CbGws1q5di82bN2PhwoUl1vXYsWO4ffs2jh49it9++w1AfhIzf/58XL9+HQcOHEB8fDx/ovf398e+ffsAAHFxcUhKSsKKFSsAADNnzsSmTZuwZs0a3Lp1C1OnTsWwYcNw6tSpEmMgpFIwed1BQkiZjBw5kolEImZvb88/Bg8ebLDsjz/+yNzd3fnnmzZtYs7OzvxzR0dHtnnzZoP7Dh8+nI0dO1Zv25kzZ5hQKGR5eXkG95k9ezYTiUTs8ePH/LZDhw4xoVDIkpKSGGOM+fr6soULF+rt9/LLL7OPP/6YMcbYxIkT2WuvvcY4jjP4HgDY/v37GWOMxcfHMwDs6tWremVGjhzJ+vfvzz/v168fGz16NP987dq1zNvbm18m9pVXXmGLFi3SO8a2bduYj4+PwRgK3sPLy4upVKpiyzDG2MWLFxkAlp2dzRj7bznYwssQ5+TkMJlMxs6dO6e375gxY9g777xT4vEJqQyoTwAhFtS1a1esWbOGf25vbw8AOHHiBBYtWoTY2FjI5XJotVoolUrk5ubyZQqLiIjA+++/j23btqFbt25466230KBBAwBATEwM7t27hx07dvDlGWPgOA7x8fFo2rSpwdjq1KmD2rVr889DQ0PBcRzi4uJgZ2eHxMREdOzYUW+fjh074vr16wDym9m7d++OwMBA9OzZE3369EGPHj3K+EnlGzp0KMaOHYvo6GhIpVLs2LEDQ4YMgUgk4ut66dIlvSt/nU4HpVIJhUIBOzs7g8dt2bJlkX4AV69exZw5c3Dt2jWkp6fztyESEhLQrFkzg8eJjY2FUqlE9+7d9bar1WoEBQWVud6EWAolAYRYkL29PRo2bKi37dGjR+jduzfGjRuH+fPnw83NDWfPnsWYMWOg0WgMHmfOnDl499138fvvv+PQoUOYPXs2du3ahYEDB4LjOHz44YeYNGlSkf3q1KljdKwCgUDvzxf/DuQnFwXbgoODER8fj0OHDuHPP//E22+/jW7dupXr/njfvn3BcRx+//13vPzyyzhz5gyWLl3Kv85xHObOnYtBgwYV2VcmkxV73BcTq9zcXPTo0QM9evTA9u3bUatWLSQkJCAsLAxqtbrY4xQkCr///jv8/Pz0XpNKpUbVkRBroiSAECu7fPkytFotlixZAqEwv5vOjz/+WOp+jRs3RuPGjTF16lS888472LRpEwYOHIjg4GDcunWrSLJRmoSEBCQmJsLX1xcAcP78eQiFQjRu3BhOTk7w9fXF2bNn0blzZ36fc+fOoW3btvxzJycnhIeHIzw8HIMHD0bPnj2Rnp4ONzc3vfcquArX6XQlxmRra4tBgwZhx44duHfvHho3boyQkBD+9eDgYMTFxZlc1xf9+++/SE1NxeLFi+Hv7w8Aeh0ti4u5WbNmkEqlSEhIQJcuXcoVAyHWQEkAIVbWoEEDaLVarFq1Cn379sVff/2F7777rtjyeXl5+PTTTzF48GAEBATgyZMnuHTpEt58800AwOeff4727dtj/Pjx+OCDD2Bvb893glu1alWxx5XJZBg5ciS+/fZbyOVyTJo0CW+//TY/tO/TTz/F7Nmz0aBBA7Ru3RqbNm3CtWvX+NsOy5Ytg4+PD1q3bg2hUIg9e/bA29vb4CRHnp6esLW1xeHDh1G7dm3IZLJihwcOHToUffv2xa1btzBs2DC917766iv06dMH/v7+eOuttyAUCnHjxg38888/WLBgQYmfe2F16tSBRCLBqlWrMG7cONy8eRPz58/XK1O3bl0IBAL89ttv6N27N2xtbeHo6Ihp06Zh6tSp4DgOnTp1glwux7lz5+Dg4KA3ioGQSsnanRIIqSle7PRW2NKlS5mPjw+ztbVlYWFhbOvWrXqd0Ap3DFSpVGzIkCHM39+fSSQS5uvryyZMmKDX6e/ixYuse/fuzMHBgdnb27NWrVoV6dRX2OzZs9lLL73EoqOjma+vL5PJZGzQoEEsPT2dL6PT6djcuXOZn58fE4vF7KWXXmKHDh3iX1+3bh1r3bo1s7e3Z05OTuz1119nV65c4V9HoY6BjDG2fv165u/vz4RCIevSpUuxn5FWq2U+Pj4MALt//36R2A8fPsw6dOjAbG1tmZOTE2vbti1bt25dsXUt7t/hhx9+YPXq1WNSqZSFhoayX375pUjnxXnz5jFvb28mEAjYyJEjGWOMcRzHVqxYwQIDA5lYLGa1atViYWFh7NSpU8XGQEhlIWCMMeumIYQQa5szZw4OHDhQZAY/Qkj1RvMEEEIIITUUJQGEEEJIDUW3AwghhJAailoCCCGEkBqKkgBCCCGkhqIkgBBCCKmhKAkghBBCaihKAgghhJAaipIAQgghpIaiJIAQQgipoSgJIIQQQmqo/wMj5sbC3+RTpAAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 1200x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from sklearn import metrics\n",
    "from tqdm import tqdm\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns; sns.set()\n",
    "plt.style.use('default')\n",
    "import matplotlib\n",
    "import pickle\n",
    "\n",
    "\n",
    "use_sigmoid = True\n",
    "\n",
    "seed_list = [1234, 1111, 2222, 3333, 4444]\n",
    "best_pt_path_list = ['ddrsa_eeg_cr3_1234_alpha1.0_beta5_gamma0.5_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230318_0133',\n",
    "                     'ddrsa_eeg_cr3_1111_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230324_0049',\n",
    "                     'ddrsa_eeg_cr3_2222_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230317_0946',\n",
    "                     'ddrsa_eeg_cr3_3333_alpha1.0_beta1_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230328_0242',\n",
    "                     'ddrsa_eeg_cr3_4444_alpha1.0_beta5_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230328_0907',\n",
    "                    ]\n",
    "\n",
    "data_mode = 'EEG_cr3' # ['EEG_nocr', 'EEG_cr2', 'EEG_cr4']\n",
    "\n",
    "if data_mode == 'EEG_cr3':\n",
    "    path = '../../eeg/competing-risk/EEG_processed_data_long_by_death_cat_exclude_3.csv'\n",
    "\n",
    "(x_dim, x_dim_cont, x_dim_bin), (data, time, label, time_original, time_to_last), \\\n",
    "    (mask1, mask2, mask3), (data_mi), trans_discrete_time = \\\n",
    "                impt.import_dataset(path)\n",
    "\n",
    "_, num_Event, num_Category  = np.shape(mask1)  # dim of mask3: [subj, Num_Event, Num_Category]\n",
    "max_length                  = np.shape(data)[1]\n",
    "\n",
    "pred_time = list(range(1,13))\n",
    "eval_time = [24, 48, 72]\n",
    "mean_fpr = np.linspace(1e-10, 1, 1000)\n",
    "\n",
    "tprs = []\n",
    "aucs = []\n",
    "\n",
    "for e, e_time in enumerate(eval_time):\n",
    "    print('eval time:', e_time, end=' ')\n",
    "    tprs = []\n",
    "    aucs = []\n",
    "    for r, seed in tqdm(enumerate(seed_list)):\n",
    "\n",
    "        best_pt_path = best_pt_path_list[r]\n",
    "\n",
    "        ### TRAINING-TESTING SPLIT\n",
    "        (tr_data,te_data, tr_data_mi, te_data_mi, tr_time,te_time, tr_label,te_label, tr_time_orinal,te_time_original,\n",
    "         tr_time_to_last,te_time_to_last,\n",
    "         tr_mask1,te_mask1, tr_mask2,te_mask2, tr_mask3,te_mask3) = \\\n",
    "                train_test_split(data, data_mi, time, label, time_original, time_to_last,\n",
    "                                 mask1, mask2, mask3, test_size=0.2, stratify=label, random_state=seed) \n",
    "\n",
    "        (tr_data,va_data, tr_data_mi, va_data_mi, tr_time,va_time, tr_label,va_label, tr_time_orinal,va_time_original,\n",
    "         tr_time_to_last, va_time_to_last,\n",
    "         tr_mask1,va_mask1, tr_mask2,va_mask2, tr_mask3,va_mask3) = \\\n",
    "                train_test_split(tr_data, tr_data_mi, tr_time, tr_label, tr_time_orinal, tr_time_to_last,\n",
    "                                 tr_mask1, tr_mask2, tr_mask3, test_size=0.2, stratify=tr_label, random_state=seed) \n",
    "\n",
    "        model_name = 'ddrsa_eeg_cr3'\n",
    "        num_Event = 3\n",
    "\n",
    "        model = \\\n",
    "            DDRSA(input_dim  = x_dim,\n",
    "                  output_dim = num_Category,\n",
    "                  layers_rnn = param['layers_rnn'],\n",
    "                  hidden_rnn = param['hidden_rnn'],\n",
    "                  long_param = {'layers': 2*[param['hidden_dim_FC']], 'dropout': dropout}, \n",
    "                  att_param  = {'layers': param['hidden_att']*[param['hidden_dim_FC']], 'dropout': dropout}, \n",
    "                  cs_param   = {'layers': param['hidden_cs']*[param['hidden_dim_FC']], 'dropout': dropout},\n",
    "                  risks      = num_Event,\n",
    "                  use_sigmoid = use_sigmoid\n",
    "                 )\n",
    "        optimizer = torch.optim.Adam(model.parameters(), lr = lr_train)\n",
    "        _ = load_checkpoint(f'checkpoints_ddrsa/{model_name}/{best_pt_path}_{seed}.pt', model, optimizer, device)\n",
    "        model = model.to(device)\n",
    "\n",
    "        prob_death_all = np.ones((te_data.shape[0], len(pred_time), len(eval_time)))\n",
    "        prob_awake_all = np.ones((te_data.shape[0], len(pred_time), len(eval_time)))\n",
    "        risk_death_old_all = np.ones((te_data.shape[0], len(pred_time), len(eval_time)))\n",
    "\n",
    "        for p, p_time in enumerate(pred_time):\n",
    "            pred_horizon = int(p_time)\n",
    "            out = _f_get_pred(model, te_data, p_time, device)\n",
    "            risk_death_awaken_all_future = 1 - out[:, :2, :pred_horizon].sum((1,2))\n",
    "\n",
    "            for t, t_time in enumerate(eval_time):\n",
    "                eval_horizon = int(t_time) + pred_horizon\n",
    "                risk_death = out[:, 0, pred_horizon:eval_horizon+1].sum(1)\n",
    "                risk_awake = out[:, 1, pred_horizon:eval_horizon+1].sum(1)\n",
    "                risk_death_old = risk_death / (risk_death+risk_awake)\n",
    "                prob_death = risk_death / risk_death_awaken_all_future\n",
    "                prob_awake = risk_awake / risk_death_awaken_all_future\n",
    "                prob_death_all[:, p, t] = prob_death\n",
    "                prob_awake_all[:, p, t] = prob_awake\n",
    "                risk_death_old_all[:, p, t] = risk_death_old\n",
    "\n",
    "        mask = (te_label==1) + (te_label==2) # for only people who died naturally, and who awoke\n",
    "\n",
    "        scores = -(prob_awake_all / prob_death_all)[:, -1, e]\n",
    "        fpr, tpr, thresholds = metrics.roc_curve(te_label[mask], scores[mask.flatten()], \n",
    "                                                 pos_label=1)\n",
    "        interp_tpr = np.interp(mean_fpr, fpr, tpr)\n",
    "        interp_tpr[0] = 0.0\n",
    "        roc_auc = metrics.auc(fpr, tpr)\n",
    "\n",
    "        tprs.append(interp_tpr)\n",
    "        aucs.append(roc_auc)\n",
    "\n",
    "    n_std = 1\n",
    "    scale = 'log' # {'log', 'linear'}\n",
    "\n",
    "    fig = plt.figure(figsize=(12,10))\n",
    "    ax = fig.add_subplot(3, 2, 1)\n",
    "    ax.plot(mean_fpr, mean_fpr, linestyle='--', color='black', alpha=.8,\n",
    "            label=r'%s (AUC = %0.3f $\\pm$ %0.3f)' % ('Random', 0, 0))\n",
    "\n",
    "    mean_tpr = np.mean(tprs, axis=0)\n",
    "    mean_tpr[-1] = 1.0\n",
    "    mean_auc = metrics.auc(mean_fpr, mean_tpr)\n",
    "    std_auc = np.std(aucs)\n",
    "    plt.plot(mean_fpr, mean_tpr, color='b',\n",
    "            label=r'%s (AUC = %0.3f $\\pm$ %0.3f)' % ('DDRSA', mean_auc, n_std*std_auc),\n",
    "            lw=2, alpha=.8)\n",
    "    print(r'%s (AUC = %0.3f ± %0.3f)' % ('DDRSA', mean_auc, n_std*std_auc))\n",
    "    std_tpr = np.std(tprs, axis=0)\n",
    "    tprs_upper = np.minimum(mean_tpr + n_std*std_tpr, 1)\n",
    "    tprs_lower = np.maximum(mean_tpr - n_std*std_tpr, 0)\n",
    "    ax.fill_between(mean_fpr, tprs_lower, tprs_upper, color='b', alpha=.2)\n",
    "\n",
    "    plt.xlim([2e-2, 5])\n",
    "    plt.yticks([0.00,0.25,0.50,0.75,1.00])\n",
    "    if scale == 'log':\n",
    "        ax.set_xscale(scale)\n",
    "        ax.set_xticks([0.03, 0.06, 0.10, 0.18, 0.32, 0.56, 1.00])\n",
    "        ax.get_xaxis().set_major_formatter(matplotlib.ticker.ScalarFormatter())\n",
    "\n",
    "    ax.set(xlim=[-0.05, 1.05], ylim=[-0.05, 1.05],\n",
    "           # title=r'Mean ROC with $\\pm$ %0.1f std. dev.' % (n_std)\n",
    "          )\n",
    "    ax.legend(loc='lower right')\n",
    "    plt.xlabel('False positive rate')\n",
    "    plt.ylabel('True positive rate')\n",
    "    \n",
    "    auc_info = {'mean_tpr': mean_tpr, 'mean_auc': mean_auc, 'std_auc': std_auc, 'tprs': tprs}\n",
    "    with open(f'ddrsa_auc_info_{12}_{e_time}.pickle', 'wb') as handle:\n",
    "        pickle.dump(auc_info,\n",
    "                    handle, protocol=pickle.HIGHEST_PROTOCOL)\n",
    "    \n",
    "    fname = r'ROC_%s_%d_%d.pdf' % ('DDRSA', 12, e_time)\n",
    "    plt.savefig(fname, dpi=200, format='pdf', bbox_inches='tight')\n",
    "    # plt.show()\n",
    "   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c27c1b88-7bcb-43f1-b8fa-53280a835c35",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "183aba61",
   "metadata": {},
   "source": [
    "# Part III: Test the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "d84fb5ef-e1b9-44e3-871c-563a79f4fd7e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-09T05:29:14.051339Z",
     "iopub.status.busy": "2023-04-09T05:29:14.050761Z",
     "iopub.status.idle": "2023-04-09T05:29:14.092829Z",
     "shell.execute_reply": "2023-04-09T05:29:14.081513Z",
     "shell.execute_reply.started": "2023-04-09T05:29:14.051288Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "def eval_model(model, te_data, pred_time, eval_time, num_Event, device='cuda'):\n",
    "    # pred_time = [4, 8, 12]\n",
    "    # eval_time = [12, 24, 36, 48, 60, 72]\n",
    "\n",
    "    risk_all = f_get_risk_predictions(model, te_data, pred_time, eval_time, device=device)\n",
    "\n",
    "    for p, p_time in enumerate(pred_time):\n",
    "        pred_horizon = int(p_time)\n",
    "        result1, result2 = np.zeros([num_Event, len(eval_time)]), np.zeros([num_Event, len(eval_time)])\n",
    "\n",
    "        for t, t_time in enumerate(eval_time):                \n",
    "            eval_horizon = int(t_time) + pred_horizon\n",
    "            for k in range(num_Event):\n",
    "                result1[k, t] = c_index(risk_all[k][:, p, t], te_time, (te_label[:,0] == k+1).astype(int), eval_horizon) #-1 for no event (not comparable)\n",
    "                result2[k, t] = brier_score(risk_all[k][:, p, t], te_time, (te_label[:,0] == k+1).astype(int), eval_horizon) #-1 for no event (not comparable)\n",
    "\n",
    "        if p == 0:\n",
    "            final1, final2 = result1, result2\n",
    "        else:\n",
    "            final1, final2 = np.append(final1, result1, axis=0), np.append(final2, result2, axis=0)\n",
    "\n",
    "\n",
    "    row_header = []\n",
    "    for p_time in pred_time:\n",
    "        for k in range(num_Event):\n",
    "            row_header.append('pred_time {}: event_{}'.format(p_time, k+1))\n",
    "\n",
    "    col_header = []\n",
    "    for t_time in eval_time:\n",
    "        col_header.append('eval_time {}'.format(t_time))\n",
    "\n",
    "    # c-index result\n",
    "    df1 = pd.DataFrame(final1, index = row_header, columns=col_header)\n",
    "\n",
    "    # brier-score result\n",
    "    df2 = pd.DataFrame(final2, index = row_header, columns=col_header)\n",
    "\n",
    "    ### PRINT RESULTS\n",
    "    print('========================================================')\n",
    "    print('--------------------------------------------------------')\n",
    "    print('- C-INDEX: ')\n",
    "    print(df1)\n",
    "    print('--------------------------------------------------------')\n",
    "    print('- BRIER-SCORE: ')\n",
    "    print(df2)\n",
    "    print('========================================================')\n",
    "    \n",
    "    return df1, df2"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "394aa46b-2442-47cc-b3b4-e57b7b327346",
   "metadata": {},
   "source": [
    "### three competing risk\n",
    "- 1: Death (Rearrested, intractable shock, multisystem organ failure; Brain death)\n",
    "- 2: Awake\n",
    "- 3: Withdrawal for perceived poor neurological prognosis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "d6fdd069-6462-4e13-8b8d-09e4b564c31f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-09T05:29:15.653146Z",
     "iopub.status.busy": "2023-04-09T05:29:15.651533Z",
     "iopub.status.idle": "2023-04-09T05:29:21.811645Z",
     "shell.execute_reply": "2023-04-09T05:29:21.810249Z",
     "shell.execute_reply.started": "2023-04-09T05:29:15.653076Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/heinz-georgenas/anaconda3/envs/py39xiaobin/lib/python3.9/site-packages/pycox/preprocessing/label_transforms.py:113: UserWarning: Calling fit method, when 'cuts' are already defined. Leaving cuts unchanged.\n",
      "  warnings.warn(\"Calling fit method, when 'cuts' are already defined. Leaving cuts unchanged.\")\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "3"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "seed = SEED\n",
    "\n",
    "data_mode = 'EEG_cr3' # ['EEG_nocr', 'EEG_cr2', 'EEG_cr4']\n",
    "\n",
    "if data_mode == 'EEG_nocr':\n",
    "    path = '../../eeg/varying-seq-len/EEG_processed_data_long.csv'\n",
    "elif data_mode == 'EEG_cr2':\n",
    "    path = '../../eeg/competing-risk/EEG_processed_data_long.csv'\n",
    "elif data_mode == 'EEG_cr4':\n",
    "    path = '../../eeg/competing-risk/EEG_processed_data_long_by_death_cat.csv'\n",
    "elif data_mode == 'EEG_cr3':\n",
    "    path = '../../eeg/competing-risk/EEG_processed_data_long_by_death_cat_exclude_3.csv'\n",
    "\n",
    "(x_dim, x_dim_cont, x_dim_bin), (data, time, label, time_original, time_to_last), \\\n",
    "    (mask1, mask2, mask3), (data_mi), trans_discrete_time = \\\n",
    "                impt.import_dataset(path)\n",
    "\n",
    "_, num_Event, num_Category  = np.shape(mask1)  # dim of mask3: [subj, Num_Event, Num_Category]\n",
    "max_length                  = np.shape(data)[1]\n",
    "\n",
    "### TRAINING-TESTING SPLIT\n",
    "(tr_data,te_data, tr_data_mi, te_data_mi, tr_time,te_time, tr_label,te_label, tr_time_orinal,te_time_original,\n",
    " tr_time_to_last,te_time_to_last,\n",
    " tr_mask1,te_mask1, tr_mask2,te_mask2, tr_mask3,te_mask3) = \\\n",
    "        train_test_split(data, data_mi, time, label, time_original, time_to_last,\n",
    "                         mask1, mask2, mask3, test_size=0.2, stratify=label, random_state=seed) \n",
    "\n",
    "(tr_data,va_data, tr_data_mi, va_data_mi, tr_time,va_time, tr_label,va_label, tr_time_orinal,va_time_original,\n",
    " tr_time_to_last, va_time_to_last,\n",
    " tr_mask1,va_mask1, tr_mask2,va_mask2, tr_mask3,va_mask3) = \\\n",
    "        train_test_split(tr_data, tr_data_mi, tr_time, tr_label, tr_time_orinal, tr_time_to_last,\n",
    "                         tr_mask1, tr_mask2, tr_mask3, test_size=0.2, stratify=tr_label, random_state=seed) \n",
    "\n",
    "num_Event"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "fd092867-640a-4af8-8459-8d7ffefc0310",
   "metadata": {
    "tags": []
   },
   "source": [
    "#### DDRSA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "e8406f08-f268-4bd9-88e2-57f80c2f5081",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-03T05:18:18.909205Z",
     "iopub.status.busy": "2023-04-03T05:18:18.908784Z",
     "iopub.status.idle": "2023-04-03T05:18:19.465983Z",
     "shell.execute_reply": "2023-04-03T05:18:19.465340Z",
     "shell.execute_reply.started": "2023-04-03T05:18:18.909167Z"
    },
    "scrolled": true,
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "========================================================\n",
      "--------------------------------------------------------\n",
      "- C-INDEX: \n",
      "                       eval_time 24  eval_time 48  eval_time 72  eval_time 96\n",
      "pred_time 6: event_1       0.671414      0.648041      0.689655      0.724554\n",
      "pred_time 6: event_2       0.803154      0.847145      0.838727      0.821962\n",
      "pred_time 6: event_3       0.617995      0.545581      0.603801      0.597442\n",
      "pred_time 12: event_1      0.759639      0.715438      0.729443      0.727899\n",
      "pred_time 12: event_2      0.799552      0.787920      0.791343      0.769490\n",
      "pred_time 12: event_3      0.536557      0.577127      0.607293      0.605910\n",
      "--------------------------------------------------------\n",
      "- BRIER-SCORE: \n",
      "                       eval_time 24  eval_time 48  eval_time 72  eval_time 96\n",
      "pred_time 6: event_1       0.038770      0.078816      0.107982      0.122266\n",
      "pred_time 6: event_2       0.052299      0.099254      0.127955      0.142834\n",
      "pred_time 6: event_3       0.083791      0.154161      0.190634      0.205454\n",
      "pred_time 12: event_1      0.046571      0.085154      0.109830      0.128017\n",
      "pred_time 12: event_2      0.062801      0.108201      0.133143      0.151935\n",
      "pred_time 12: event_3      0.099585      0.165574      0.195419      0.210678\n",
      "========================================================\n"
     ]
    }
   ],
   "source": [
    "pred_time = [6,12]\n",
    "eval_time = [24, 48, 72, 96]\n",
    "model_name = 'ddrsa_eeg_cr3'\n",
    "\n",
    "use_sigmoid = False\n",
    "hidden_att  = 3\n",
    "hidden_cs   = 3\n",
    "\n",
    "best_pt_path = 'ddrsa_eeg_cr3_1234_alpha1.0_beta5_gamma0.5_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230318_0133'\n",
    "best_pt_path = 'ddrsa_eeg_cr3_1234_sigmoidFalse_alpha1.0_beta0.5_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt3_hiddenCS3_dropout0.4_20230323_1331'\n",
    "# best_pt_path = 'ddrsa_eeg_cr3_1111_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230324_0049'\n",
    "# best_pt_path = 'ddrsa_eeg_cr3_2222_alpha1.0_beta1_gamma0.1_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230317_0946'\n",
    "# best_pt_path = 'ddrsa_eeg_cr3_3333_alpha1.0_beta1_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.4_20230328_0242'\n",
    "# best_pt_path = 'ddrsa_eeg_cr3_4444_alpha1.0_beta5_gamma0.05_BSZ32_Lr0.0001_hiddenRNN64_hiddenFC64_layersRNN2_hiddenAtt2_hiddenCS2_dropout0.2_20230328_0907'\n",
    "model = \\\n",
    "    DDRSA(input_dim  = x_dim,\n",
    "              output_dim = num_Category,\n",
    "              layers_rnn = param['layers_rnn'],\n",
    "              hidden_rnn = param['hidden_rnn'],\n",
    "              long_param = {'layers': 2*[param['hidden_dim_FC']], 'dropout': dropout}, \n",
    "              att_param  = {'layers': 3*[param['hidden_dim_FC']], 'dropout': dropout}, \n",
    "              cs_param   = {'layers': 3*[param['hidden_dim_FC']], 'dropout': dropout},\n",
    "              risks      = num_Event,\n",
    "              use_sigmoid = use_sigmoid\n",
    "             )\n",
    "optimizer = torch.optim.Adam(model.parameters(), lr = lr_train)\n",
    "_ = load_checkpoint(f'checkpoints_ddrsa/{model_name}/{best_pt_path}_{SEED}.pt', model, optimizer, device)\n",
    "model = model.to(device)\n",
    "\n",
    "df1, df2 = eval_model(model, te_data, pred_time, eval_time, num_Event)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b05a932d-9f69-4290-8563-42789d030af5",
   "metadata": {},
   "source": [
    "### c-index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "d46301c3-89c3-4900-9745-c208410520e9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-03T05:18:25.439017Z",
     "iopub.status.busy": "2023-04-03T05:18:25.438019Z",
     "iopub.status.idle": "2023-04-03T05:18:25.452043Z",
     "shell.execute_reply": "2023-04-03T05:18:25.451333Z",
     "shell.execute_reply.started": "2023-04-03T05:18:25.438939Z"
    },
    "tags": []
   },
   "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>eval_time 24</th>\n",
       "      <th>eval_time 48</th>\n",
       "      <th>eval_time 72</th>\n",
       "      <th>eval_time 96</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>pred_time 6: event_1</th>\n",
       "      <td>0.671414</td>\n",
       "      <td>0.648041</td>\n",
       "      <td>0.689655</td>\n",
       "      <td>0.724554</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pred_time 12: event_1</th>\n",
       "      <td>0.759639</td>\n",
       "      <td>0.715438</td>\n",
       "      <td>0.729443</td>\n",
       "      <td>0.727899</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       eval_time 24  eval_time 48  eval_time 72  eval_time 96\n",
       "pred_time 6: event_1       0.671414      0.648041      0.689655      0.724554\n",
       "pred_time 12: event_1      0.759639      0.715438      0.729443      0.727899"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1[::3] # 1: death"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "b748497a-2baa-4e65-8a2a-b4c029c5728e",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-03T05:18:25.756253Z",
     "iopub.status.busy": "2023-04-03T05:18:25.755392Z",
     "iopub.status.idle": "2023-04-03T05:18:25.778750Z",
     "shell.execute_reply": "2023-04-03T05:18:25.777777Z",
     "shell.execute_reply.started": "2023-04-03T05:18:25.756173Z"
    },
    "tags": []
   },
   "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>eval_time 24</th>\n",
       "      <th>eval_time 48</th>\n",
       "      <th>eval_time 72</th>\n",
       "      <th>eval_time 96</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>pred_time 6: event_2</th>\n",
       "      <td>0.803154</td>\n",
       "      <td>0.847145</td>\n",
       "      <td>0.838727</td>\n",
       "      <td>0.821962</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pred_time 12: event_2</th>\n",
       "      <td>0.799552</td>\n",
       "      <td>0.787920</td>\n",
       "      <td>0.791343</td>\n",
       "      <td>0.769490</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       eval_time 24  eval_time 48  eval_time 72  eval_time 96\n",
       "pred_time 6: event_2       0.803154      0.847145      0.838727      0.821962\n",
       "pred_time 12: event_2      0.799552      0.787920      0.791343      0.769490"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1[1::3] # 2: awake"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "2a22f902-223b-423a-b82e-17292b523c1d",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2023-04-03T05:18:26.101371Z",
     "iopub.status.busy": "2023-04-03T05:18:26.100541Z",
     "iopub.status.idle": "2023-04-03T05:18:26.124691Z",
     "shell.execute_reply": "2023-04-03T05:18:26.122829Z",
     "shell.execute_reply.started": "2023-04-03T05:18:26.101293Z"
    },
    "tags": []
   },
   "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>eval_time 24</th>\n",
       "      <th>eval_time 48</th>\n",
       "      <th>eval_time 72</th>\n",
       "      <th>eval_time 96</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>pred_time 6: event_3</th>\n",
       "      <td>0.617995</td>\n",
       "      <td>0.545581</td>\n",
       "      <td>0.603801</td>\n",
       "      <td>0.597442</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>pred_time 12: event_3</th>\n",
       "      <td>0.536557</td>\n",
       "      <td>0.577127</td>\n",
       "      <td>0.607293</td>\n",
       "      <td>0.605910</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                       eval_time 24  eval_time 48  eval_time 72  eval_time 96\n",
       "pred_time 6: event_3       0.617995      0.545581      0.603801      0.597442\n",
       "pred_time 12: event_3      0.536557      0.577127      0.607293      0.605910"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df1[2::3] # 3: Withdrawal for perceived poor neurological prognosis"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "78ede6ff-d43a-40cc-8935-89faddb3e5e7",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {
    "height": "calc(100% - 180px)",
    "left": "10px",
    "top": "150px",
    "width": "349.091px"
   },
   "toc_section_display": true,
   "toc_window_display": true
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}