k-fold-cox-time-survival-best.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd\n",
    "# For preprocessing\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn_pandas import DataFrameMapper \n",
    "from pycox.evaluation import EvalSurv\n",
    "\n",
    "from pycox.models import CoxTime\n",
    "import torch # For building the networks \n",
    "from sklearn.model_selection import KFold, GridSearchCV\n",
    "from estimators import CoxTimeEstimator\n",
    "import torchtuples as tt \n",
    "from collections import defaultdict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Hyper-parameters\n",
    "test_size = 0.2\n",
    "train_size = 0.8\n",
    "assert test_size + train_size == 1, \"total size is not equal to 1\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# We also set some seeds to make this reproducable.\n",
    "# Note that on gpu, there is still some randomness.\n",
    "seed = 1234\n",
    "np.random.seed(seed)\n",
    "torch.manual_seed(seed);"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dataset\n",
    "\n",
    "We load data from an Iranian Institution\n",
    "\n",
    "The `time` column gives the observed times and the `event` column contains indicators of whether the observation is an event (1) or a censored observation (0)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>event</th>\n",
       "      <th>time</th>\n",
       "      <th>diagnosic-age</th>\n",
       "      <th>edjucation</th>\n",
       "      <th>marital-status</th>\n",
       "      <th>G</th>\n",
       "      <th>AB</th>\n",
       "      <th>BF</th>\n",
       "      <th>FH</th>\n",
       "      <th>Surgery</th>\n",
       "      <th>T</th>\n",
       "      <th>N</th>\n",
       "      <th>stage</th>\n",
       "      <th>Path</th>\n",
       "      <th>Grade</th>\n",
       "      <th>LVI</th>\n",
       "      <th>hormone subgroup</th>\n",
       "      <th>CT</th>\n",
       "      <th>RT</th>\n",
       "      <th>HT</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>334.2</td>\n",
       "      <td>46</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0</td>\n",
       "      <td>328.9</td>\n",
       "      <td>35</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>4</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0</td>\n",
       "      <td>310.4</td>\n",
       "      <td>41</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0</td>\n",
       "      <td>309.7</td>\n",
       "      <td>48</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>299.6</td>\n",
       "      <td>47</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   event   time  diagnosic-age  edjucation  marital-status  G  AB  BF  FH  \\\n",
       "0      0  334.2             46           2               2  2   1   1   0   \n",
       "1      0  328.9             35           1               2  2   0   1   0   \n",
       "2      0  310.4             41           2               2  2   0   1   0   \n",
       "3      0  309.7             48           2               2  1   0   1   0   \n",
       "4      1  299.6             47           4               2  3   0   2   2   \n",
       "\n",
       "   Surgery  T  N  stage  Path  Grade  LVI  hormone subgroup  CT  RT  HT  \n",
       "0        2  1  2      3     1      2    0                 2   1   1   1  \n",
       "1        1  3  0      2     1      2    0                 4   1   1   1  \n",
       "2        2  1  2      3     1      2    1                 2   1   1   1  \n",
       "3        2  2  0      2     1      2    0                 2   1   0   1  \n",
       "4        2  2  0      2     1      2    0                 2   1   1   1  "
      ]
     },
     "execution_count": 26,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_dataset_fp ='finaldata-June26.csv'\n",
    "df = pd.read_csv(train_dataset_fp)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x864 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "\n",
    "s = sns.displot(df, x=\"time\", hue=\"hormone subgroup\", multiple=\"stack\")\n",
    "s.set(xlabel='time(months)', ylabel='number of patients')\n",
    "plt.text(251+0.5,250, \"Luminal B\", horizontalalignment='left', size='medium', color='black', weight='normal')\n",
    "plt.text(251+0.5,242, \"Luminal A\", horizontalalignment='left', size='medium', color='black', weight='normal')\n",
    "plt.text(251+0.5,234, \"HER2-Enriched\", horizontalalignment='left', size='medium', color='black', weight='normal')\n",
    "plt.text(251+0.5,226, \"Triple-Negative\", horizontalalignment='left', size='medium', color='black', weight='normal')\n",
    "plt.gcf().set_size_inches(15, 12)\n",
    "sns.move_legend(s, \"center\", bbox_to_anchor=(.66, .66))  "
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train-Test-Spilit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train, df_test = train_test_split(df, test_size=test_size, random_state=seed, stratify=df[\"hormone subgroup\"])"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Feature transforms\n",
    "\n",
    "The dataset has 18 covariates.\n",
    "We will standardize the 12 categorical covariates, and leave the binary covariates as is.\n",
    "Note that PyTorch require variables of type `'float32'`.\n",
    "\n",
    "Here we use the `sklearn_pandas.DataFrameMapper` to make feature mappers, but this has nothing to do the the `pycox` package."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols_standardize = ['diagnosic-age',\n",
    "                    'edjucation', \n",
    "                    'G',\n",
    "                    'BF',\n",
    "                    'FH',\n",
    "                    'T',\n",
    "                    'N',\n",
    "                    'stage',\n",
    "                    'Grade',\n",
    "                    'CT','RT','hormone subgroup']\n",
    "cols_leave = ['marital-status', 'AB', 'Path', 'LVI', 'Surgery', 'HT']\n",
    "\n",
    "standardize = [([col], StandardScaler()) for col in cols_standardize]\n",
    "leave = [(col, None) for col in cols_leave]\n",
    "\n",
    "x_mapper = DataFrameMapper(standardize + leave)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# apply scalar transformation!\n",
    "x_train = x_mapper.fit_transform(df_train).astype('float32')\n",
    "x_test = x_mapper.transform(df_test).astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[(1, 101), (2, 435), (3, 54), (4, 126)]\n",
      "[(1, 402), (2, 1740), (3, 216), (4, 506)]\n"
     ]
    }
   ],
   "source": [
    "# Checking the distribution of `hormone subgroup`\n",
    "feature_index = 11 # hormone subgroup\n",
    "\n",
    "def get_feature_distribution(feature_index, dataset):\n",
    "    filtered_individuals = defaultdict(list)\n",
    "    dataset_reverse = standardize[feature_index][1].inverse_transform(dataset)\n",
    "\n",
    "    for i in range(len(dataset)):\n",
    "        individual = dataset_reverse[i]\n",
    "        filtered_individuals[int(individual[feature_index])].append(i)\n",
    "    print(sorted([(k, len(v)) for k, v in filtered_individuals.items()]))\n",
    "    return filtered_individuals\n",
    "get_feature_distribution(feature_index, x_test);\n",
    "get_feature_distribution(feature_index, x_train);"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Label transforms\n",
    "\n",
    "The survival methods require individual label transforms, so we have included a proposed `label_transform` for each method.\n",
    "For most of them the `label_transform` is just a shorthand for the class `pycox.preprocessing.label_transforms.LabTransDiscreteTime`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([0, 1], dtype=int64), array([3146,  434], dtype=int64))"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# events!\n",
    "np.unique(df['event'], return_counts=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Saba\\anaconda3\\lib\\site-packages\\sklearn\\utils\\validation.py:67: FutureWarning: Pass copy=True, with_mean=True, with_std=True as keyword args. From version 0.25 passing these as positional arguments will result in an error\n",
      "  warnings.warn(\"Pass {} as keyword args. From version 0.25 \"\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<pycox.preprocessing.label_transforms.LabTransCoxTime at 0x2348cd84400>"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "label_translator = CoxTime.label_transform()\n",
    "get_features = lambda df: (df['time'].values, df['event'].values)\n",
    "label_translator.fit(*get_features(df_train))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# K-Fold"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [],
   "source": [
    "params = {  # search params\n",
    "        'learning_rate': [0.1, 0.01, 0.001],\n",
    "        'num_nodes': [[32, 32], [32, 64], [32, 64, 128], [32, 64, 128, 256]],\n",
    "        'dropout': [0.0, 0.1, 0.2, 0.3],\n",
    "        'batch_size': [256],\n",
    "    }\n",
    "\n",
    "cv = 5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "estimator = CoxTimeEstimator(label_translator, x_train.shape[1])\n",
    "gs = GridSearchCV(estimator, params, refit=True, cv=KFold(n_splits=cv), verbose=1, n_jobs=5)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "stack = lambda duration, event:np.hstack([duration.reshape((-1, 1)), event.reshape((-1, 1))])\n",
    "y_train = stack(*get_features(df_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Fitting 5 folds for each of 48 candidates, totalling 240 fits\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[Parallel(n_jobs=5)]: Using backend LokyBackend with 5 concurrent workers.\n",
      "[Parallel(n_jobs=5)]: Done  40 tasks      | elapsed:  1.1min\n",
      "[Parallel(n_jobs=5)]: Done 190 tasks      | elapsed:  4.9min\n",
      "[Parallel(n_jobs=5)]: Done 240 out of 240 | elapsed:  6.2min finished\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "GridSearchCV(cv=KFold(n_splits=5, random_state=None, shuffle=False),\n",
       "             estimator=CoxTimeEstimator(in_features=18,\n",
       "                                        label_translator=<pycox.preprocessing.label_transforms.LabTransCoxTime object at 0x000002348CD84400>),\n",
       "             n_jobs=5,\n",
       "             param_grid={'batch_size': [256], 'dropout': [0.0, 0.1, 0.2, 0.3],\n",
       "                         'learning_rate': [0.1, 0.01, 0.001],\n",
       "                         'num_nodes': [[32, 32], [32, 64], [32, 64, 128],\n",
       "                                       [32, 64, 128, 256]]},\n",
       "             verbose=1)"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "gs.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.7829686397949981 {'batch_size': 256, 'dropout': 0.1, 'learning_rate': 0.1, 'num_nodes': [32, 32]}\n"
     ]
    }
   ],
   "source": [
    "print(gs.best_score_, gs.best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7358386944923898"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "durations_test, events_test = get_features(df_test)\n",
    "gs.score(x_test, np.hstack([durations_test.reshape((-1, 1)), events_test.reshape((-1, 1))]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "estimator = CoxTimeEstimator(label_translator, x_train.shape[1], **gs.best_params_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "CoxTimeEstimator(dropout=0.1, in_features=18,\n",
       "                 label_translator=<pycox.preprocessing.label_transforms.LabTransCoxTime object at 0x000002348CD84400>,\n",
       "                 learning_rate=0.1)"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator.fit(x_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7243841205083947"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "estimator.score_split(x_test, durations_test, events_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "MLPVanillaCoxTime(\n",
       "  (net): MLPVanilla(\n",
       "    (net): Sequential(\n",
       "      (0): DenseVanillaBlock(\n",
       "        (linear): Linear(in_features=19, out_features=32, bias=True)\n",
       "        (activation): ReLU()\n",
       "        (batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (dropout): Dropout(p=0.1, inplace=False)\n",
       "      )\n",
       "      (1): DenseVanillaBlock(\n",
       "        (linear): Linear(in_features=32, out_features=32, bias=True)\n",
       "        (activation): ReLU()\n",
       "        (batch_norm): BatchNorm1d(32, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)\n",
       "        (dropout): Dropout(p=0.1, inplace=False)\n",
       "      )\n",
       "      (2): Linear(in_features=32, out_features=1, bias=False)\n",
       "    )\n",
       "  )\n",
       ")"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = estimator.model\n",
    "log = model.log\n",
    "model.net"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "log.plot();"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, we set the `batch_size` and the number of training `epochs`.\n",
    "\n",
    "We also include the `EarlyStopping` callback to stop training when the validation loss stops improving."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.27303454279899597"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "log.to_pandas().val_loss.min()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Prediction\n",
    "\n",
    "For evaluation we first need to obtain survival estimates for the test set.\n",
    "This can be done with `model.predict_surv` which returns an array of survival estimates, or with `model.predict_surv_df` which returns the survival estimates as a dataframe."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>706</th>\n",
       "      <th>707</th>\n",
       "      <th>708</th>\n",
       "      <th>709</th>\n",
       "      <th>710</th>\n",
       "      <th>711</th>\n",
       "      <th>712</th>\n",
       "      <th>713</th>\n",
       "      <th>714</th>\n",
       "      <th>715</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1.6</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2.0</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>...</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000e+00</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2.3</th>\n",
       "      <td>0.999999</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999922</td>\n",
       "      <td>0.999975</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999996</td>\n",
       "      <td>0.998127</td>\n",
       "      <td>0.999981</td>\n",
       "      <td>0.999998</td>\n",
       "      <td>0.999940</td>\n",
       "      <td>...</td>\n",
       "      <td>0.999995</td>\n",
       "      <td>0.999218</td>\n",
       "      <td>0.999930</td>\n",
       "      <td>0.999996</td>\n",
       "      <td>0.999988</td>\n",
       "      <td>0.999999</td>\n",
       "      <td>9.964924e-01</td>\n",
       "      <td>0.999999</td>\n",
       "      <td>0.999683</td>\n",
       "      <td>0.999999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3.0</th>\n",
       "      <td>0.999998</td>\n",
       "      <td>0.999999</td>\n",
       "      <td>0.999843</td>\n",
       "      <td>0.999950</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999992</td>\n",
       "      <td>0.996255</td>\n",
       "      <td>0.999963</td>\n",
       "      <td>0.999996</td>\n",
       "      <td>0.999879</td>\n",
       "      <td>...</td>\n",
       "      <td>0.999990</td>\n",
       "      <td>0.998435</td>\n",
       "      <td>0.999860</td>\n",
       "      <td>0.999993</td>\n",
       "      <td>0.999977</td>\n",
       "      <td>0.999997</td>\n",
       "      <td>9.929961e-01</td>\n",
       "      <td>0.999999</td>\n",
       "      <td>0.999365</td>\n",
       "      <td>0.999999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3.3</th>\n",
       "      <td>0.999998</td>\n",
       "      <td>0.999999</td>\n",
       "      <td>0.999843</td>\n",
       "      <td>0.999950</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.999992</td>\n",
       "      <td>0.996255</td>\n",
       "      <td>0.999963</td>\n",
       "      <td>0.999996</td>\n",
       "      <td>0.999879</td>\n",
       "      <td>...</td>\n",
       "      <td>0.999990</td>\n",
       "      <td>0.998435</td>\n",
       "      <td>0.999860</td>\n",
       "      <td>0.999993</td>\n",
       "      <td>0.999977</td>\n",
       "      <td>0.999997</td>\n",
       "      <td>9.929961e-01</td>\n",
       "      <td>0.999999</td>\n",
       "      <td>0.999365</td>\n",
       "      <td>0.999999</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>293.7</th>\n",
       "      <td>0.930942</td>\n",
       "      <td>0.935941</td>\n",
       "      <td>0.236014</td>\n",
       "      <td>0.926456</td>\n",
       "      <td>0.976482</td>\n",
       "      <td>0.849514</td>\n",
       "      <td>0.000385</td>\n",
       "      <td>0.693978</td>\n",
       "      <td>0.862893</td>\n",
       "      <td>0.794195</td>\n",
       "      <td>...</td>\n",
       "      <td>0.943418</td>\n",
       "      <td>0.002930</td>\n",
       "      <td>0.103643</td>\n",
       "      <td>0.769835</td>\n",
       "      <td>0.778413</td>\n",
       "      <td>0.883004</td>\n",
       "      <td>4.883350e-09</td>\n",
       "      <td>0.939780</td>\n",
       "      <td>0.059119</td>\n",
       "      <td>0.956252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>294.0</th>\n",
       "      <td>0.930942</td>\n",
       "      <td>0.935941</td>\n",
       "      <td>0.236014</td>\n",
       "      <td>0.926456</td>\n",
       "      <td>0.976482</td>\n",
       "      <td>0.849514</td>\n",
       "      <td>0.000385</td>\n",
       "      <td>0.693978</td>\n",
       "      <td>0.862893</td>\n",
       "      <td>0.794195</td>\n",
       "      <td>...</td>\n",
       "      <td>0.943418</td>\n",
       "      <td>0.002930</td>\n",
       "      <td>0.103643</td>\n",
       "      <td>0.769835</td>\n",
       "      <td>0.778413</td>\n",
       "      <td>0.883004</td>\n",
       "      <td>4.883350e-09</td>\n",
       "      <td>0.939780</td>\n",
       "      <td>0.059119</td>\n",
       "      <td>0.956252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>309.7</th>\n",
       "      <td>0.930942</td>\n",
       "      <td>0.935941</td>\n",
       "      <td>0.236014</td>\n",
       "      <td>0.926456</td>\n",
       "      <td>0.976482</td>\n",
       "      <td>0.849514</td>\n",
       "      <td>0.000385</td>\n",
       "      <td>0.693978</td>\n",
       "      <td>0.862893</td>\n",
       "      <td>0.794195</td>\n",
       "      <td>...</td>\n",
       "      <td>0.943418</td>\n",
       "      <td>0.002930</td>\n",
       "      <td>0.103643</td>\n",
       "      <td>0.769835</td>\n",
       "      <td>0.778413</td>\n",
       "      <td>0.883004</td>\n",
       "      <td>4.883350e-09</td>\n",
       "      <td>0.939780</td>\n",
       "      <td>0.059119</td>\n",
       "      <td>0.956252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>328.9</th>\n",
       "      <td>0.930942</td>\n",
       "      <td>0.935941</td>\n",
       "      <td>0.236014</td>\n",
       "      <td>0.926456</td>\n",
       "      <td>0.976482</td>\n",
       "      <td>0.849514</td>\n",
       "      <td>0.000385</td>\n",
       "      <td>0.693978</td>\n",
       "      <td>0.862893</td>\n",
       "      <td>0.794195</td>\n",
       "      <td>...</td>\n",
       "      <td>0.943418</td>\n",
       "      <td>0.002930</td>\n",
       "      <td>0.103643</td>\n",
       "      <td>0.769835</td>\n",
       "      <td>0.778413</td>\n",
       "      <td>0.883004</td>\n",
       "      <td>4.883350e-09</td>\n",
       "      <td>0.939780</td>\n",
       "      <td>0.059119</td>\n",
       "      <td>0.956252</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>334.2</th>\n",
       "      <td>0.930942</td>\n",
       "      <td>0.935941</td>\n",
       "      <td>0.236014</td>\n",
       "      <td>0.926456</td>\n",
       "      <td>0.976482</td>\n",
       "      <td>0.849514</td>\n",
       "      <td>0.000385</td>\n",
       "      <td>0.693978</td>\n",
       "      <td>0.862893</td>\n",
       "      <td>0.794195</td>\n",
       "      <td>...</td>\n",
       "      <td>0.943418</td>\n",
       "      <td>0.002930</td>\n",
       "      <td>0.103643</td>\n",
       "      <td>0.769835</td>\n",
       "      <td>0.778413</td>\n",
       "      <td>0.883004</td>\n",
       "      <td>4.883350e-09</td>\n",
       "      <td>0.939780</td>\n",
       "      <td>0.059119</td>\n",
       "      <td>0.956252</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1224 rows × 716 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "            0         1         2         3         4         5         6    \\\n",
       "1.6    1.000000  1.000000  1.000000  1.000000  1.000000  1.000000  1.000000   \n",
       "2.0    1.000000  1.000000  1.000000  1.000000  1.000000  1.000000  1.000000   \n",
       "2.3    0.999999  1.000000  0.999922  0.999975  1.000000  0.999996  0.998127   \n",
       "3.0    0.999998  0.999999  0.999843  0.999950  1.000000  0.999992  0.996255   \n",
       "3.3    0.999998  0.999999  0.999843  0.999950  1.000000  0.999992  0.996255   \n",
       "...         ...       ...       ...       ...       ...       ...       ...   \n",
       "293.7  0.930942  0.935941  0.236014  0.926456  0.976482  0.849514  0.000385   \n",
       "294.0  0.930942  0.935941  0.236014  0.926456  0.976482  0.849514  0.000385   \n",
       "309.7  0.930942  0.935941  0.236014  0.926456  0.976482  0.849514  0.000385   \n",
       "328.9  0.930942  0.935941  0.236014  0.926456  0.976482  0.849514  0.000385   \n",
       "334.2  0.930942  0.935941  0.236014  0.926456  0.976482  0.849514  0.000385   \n",
       "\n",
       "            7         8         9    ...       706       707       708  \\\n",
       "1.6    1.000000  1.000000  1.000000  ...  1.000000  1.000000  1.000000   \n",
       "2.0    1.000000  1.000000  1.000000  ...  1.000000  1.000000  1.000000   \n",
       "2.3    0.999981  0.999998  0.999940  ...  0.999995  0.999218  0.999930   \n",
       "3.0    0.999963  0.999996  0.999879  ...  0.999990  0.998435  0.999860   \n",
       "3.3    0.999963  0.999996  0.999879  ...  0.999990  0.998435  0.999860   \n",
       "...         ...       ...       ...  ...       ...       ...       ...   \n",
       "293.7  0.693978  0.862893  0.794195  ...  0.943418  0.002930  0.103643   \n",
       "294.0  0.693978  0.862893  0.794195  ...  0.943418  0.002930  0.103643   \n",
       "309.7  0.693978  0.862893  0.794195  ...  0.943418  0.002930  0.103643   \n",
       "328.9  0.693978  0.862893  0.794195  ...  0.943418  0.002930  0.103643   \n",
       "334.2  0.693978  0.862893  0.794195  ...  0.943418  0.002930  0.103643   \n",
       "\n",
       "            709       710       711           712       713       714  \\\n",
       "1.6    1.000000  1.000000  1.000000  1.000000e+00  1.000000  1.000000   \n",
       "2.0    1.000000  1.000000  1.000000  1.000000e+00  1.000000  1.000000   \n",
       "2.3    0.999996  0.999988  0.999999  9.964924e-01  0.999999  0.999683   \n",
       "3.0    0.999993  0.999977  0.999997  9.929961e-01  0.999999  0.999365   \n",
       "3.3    0.999993  0.999977  0.999997  9.929961e-01  0.999999  0.999365   \n",
       "...         ...       ...       ...           ...       ...       ...   \n",
       "293.7  0.769835  0.778413  0.883004  4.883350e-09  0.939780  0.059119   \n",
       "294.0  0.769835  0.778413  0.883004  4.883350e-09  0.939780  0.059119   \n",
       "309.7  0.769835  0.778413  0.883004  4.883350e-09  0.939780  0.059119   \n",
       "328.9  0.769835  0.778413  0.883004  4.883350e-09  0.939780  0.059119   \n",
       "334.2  0.769835  0.778413  0.883004  4.883350e-09  0.939780  0.059119   \n",
       "\n",
       "            715  \n",
       "1.6    1.000000  \n",
       "2.0    1.000000  \n",
       "2.3    0.999999  \n",
       "3.0    0.999999  \n",
       "3.3    0.999999  \n",
       "...         ...  \n",
       "293.7  0.956252  \n",
       "294.0  0.956252  \n",
       "309.7  0.956252  \n",
       "328.9  0.956252  \n",
       "334.2  0.956252  \n",
       "\n",
       "[1224 rows x 716 columns]"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "surv = model.predict_surv_df(x_test)\n",
    "surv\n",
    "# columns are sample indices, values are the probability of the events"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Evaluation\n",
    "\n",
    "The `EvalSurv` class contains some useful evaluation criteria for time-to-event prediction.\n",
    "We set `censor_surv = 'km'` to state that we want to use Kaplan-Meier for estimating the censoring distribution.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\pooya\\miniconda3\\envs\\torch_gpu\\lib\\site-packages\\pycox\\evaluation\\eval_surv.py:36: FutureWarning: is_monotonic is deprecated and will be removed in a future version. Use is_monotonic_increasing instead.\n",
      "  assert pd.Series(self.index_surv).is_monotonic\n"
     ]
    }
   ],
   "source": [
    "ev = EvalSurv(surv, durations_test, events_test, censor_surv='km')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Concordance\n",
    "\n",
    "We start with the event-time concordance by [Antolini et al. 2005](https://onlinelibrary.wiley.com/doi/10.1002/sim.2427)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.7504315079240546"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ev.concordance_td('adj_antolini')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Brier Score\n",
    "\n",
    "We can plot the the [IPCW Brier score](https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291097-0258%2819990915/30%2918%3A17/18%3C2529%3A%3AAID-SIM274%3E3.0.CO%3B2-5) for a given set of times.\n",
    "Here we just use 100 time-points between the min and max duration in the test set.\n",
    "Note that the score becomes unstable for the highest times. It is therefore common to disregard the rightmost part of the graph."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "time_grid = np.linspace(durations_test.min(), durations_test.max(), 100)\n",
    "ev.brier_score(time_grid).plot()\n",
    "plt.ylabel('Brier score of Nnet-survival Model')\n",
    "_ = plt.xlabel('Time(month)')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Negative binomial log-likelihood\n",
    "\n",
    "In a similar manner, we can plot the the [IPCW negative binomial log-likelihood](https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291097-0258%2819990915/30%2918%3A17/18%3C2529%3A%3AAID-SIM274%3E3.0.CO%3B2-5)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "ev.nbll(time_grid).plot()\n",
    "plt.ylabel('NBLL')\n",
    "_ = plt.xlabel('Time')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Integrated scores\n",
    "\n",
    "The two time-dependent scores above can be integrated over time to produce a single score [Graf et al. 1999](https://onlinelibrary.wiley.com/doi/abs/10.1002/%28SICI%291097-0258%2819990915/30%2918%3A17/18%3C2529%3A%3AAID-SIM274%3E3.0.CO%3B2-5). In practice this is done by numerical integration over a defined `time_grid`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.15124683918563567"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ev.integrated_brier_score(time_grid) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.46552419515825544"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ev.integrated_nbll(time_grid) "
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}