main.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Effects of Aggressive and Conservative Strategies for Mechanical Ventilation Liberation\n",
    "This project aims to use causal inference to estimate the effects of weaning patients off mechanical ventilation sooner or later than normal, following the approach described in [Kennedy, 2018](https://arxiv.org/pdf/1704.00211.pdf). In this study we consider patients who have been on mechanical ventilation for 24 hours (our start criteria)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "from pprint import pprint\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn.calibration import CalibratedClassifierCV\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.preprocessing import LabelEncoder, StandardScaler\n",
    "\n",
    "# ALL_WEANING_PREDICTORS is a list of all covariates intially identified as potentially useful in predicting probability of weaning\n",
    "# group_by_stay is a helper function that groups a dataframe by the column 'stay_id'\n",
    "from common import (\n",
    "    ALL_WEANING_PREDICTORS,\n",
    "    group_by_stay,\n",
    "    calibration_curve\n",
    ")\n",
    "\n",
    "np.random.seed(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First, we exclude variables from consideration if they are missing more than 10% of the time in post-baseline times. Post-baseline in this case means we filter out records from patients at time points before our start criteria.\n",
    "\n",
    "```\n",
    ">>> full_df = pd.read_csv('mimic_iv_cxr_reports_labeled_notext.csv')\n",
    ">>> post_baseline_df = full_df.loc[full_df['hour_baseline'] >= 1]\n",
    ">>> post_baseline_df.to_csv('post_baseline_weaning.csv')\n",
    "```"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/aman/VSCode Files/liberation-timing-paper/venv/lib/python3.9/site-packages/IPython/core/interactiveshell.py:3169: DtypeWarning: Columns (10) have mixed types.Specify dtype option on import or set low_memory=False.\n",
      "  has_raised = await self.run_ast_nodes(code_ast.body, cell_name,\n"
     ]
    }
   ],
   "source": [
    "# Load in the required datasets\n",
    "# df contains records for the duration of each stay across all stay_ids\n",
    "df = pd.read_csv('post_baseline_weaning.csv')\n",
    "# Cohort contains information about each patient, including\n",
    "# intime, outtime, deathtime, subject_id, etc\n",
    "cohort = pd.read_csv('icustays_1.0.csv')\n",
    "careunits = pd.read_csv('careunits_1.0.csv')\n",
    "admissions = pd.read_csv('admissions.csv')\n",
    "cohort = cohort.merge(careunits).merge(admissions)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Clean the main dataframe\n",
    "Overview:\n",
    " - Remove patients who have deathtimes before the baseline criteria for the study.\n",
    " - Include only the first ICU stays for each patient\n",
    " - Fill missing values for certain columns where NaN values just mean 0\n",
    " - Create a few columns to track the value of certain variables at time t-1\n",
    " - Forward fill most of the columns to remove NaN values, since in most cases NaN simply means the variable wasn't measured at that timepoint\n",
    " - Ignore columns who still have a NaN proportion > 0.1\n",
    " \n",
    "First, remove patients whose deathtimes are listed as before the baseline criteria for this study."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ventilation status being 2 or 5 indicates invasive ventilation or the presence a tracheal tube, respectively. When the status shifts away from these values, it implies the patient was extubated, so adjust control accordingly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.loc[~df['ventilation_status'].isin([2, 5]), 'control'] = 0"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our time period of interest is every timestep up until and including the start of the first liberation attempt. So, we remove all data points past weaning."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['weaning'] = 1 - df['control']\n",
    "df['is_past_wean'] = group_by_stay(df, 'weaning').apply(\n",
    "    lambda stay: stay.cumsum().cumsum() > 1\n",
    "    )\n",
    "df = df[~df['is_past_wean']]                                   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fill missing urine output and vasopressor (`rate_std`) values with 0, since missing entries for these two columns means the value is 0 in this dataset. We do this to make sure we don't accidentally exclude the columns from the dataset because their NaN proportion appears too high, when they should really be included."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "cols_to_impute = ['urine_output', 'last_urine_output', 'rate_std', 'last_rate_std']\n",
    "df.loc[:, cols_to_impute] = df[cols_to_impute].fillna(0)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, we'll create some relevant columns (mostly ones that track the 'last' version of a given variable, i.e. the value at the previous timepoint)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['any_vaso'] = df['rate_std'] > 0\n",
    "\n",
    "for col in ('opioid', 'benzo', 'propofol','dex', 'driving_pressure', 'any_vaso'):\n",
    "    df[f'last_{col}'] = group_by_stay(df, col).shift(periods=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Remove any stays corresponding to patients whose deathtimes occured before the baseline time for the study."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.merge(cohort[['stay_id', 'deathtime', 'intime']])\n",
    "# Find the actual time at which the patient's stay started,\n",
    "# rather than hour `44`, for example\n",
    "df['baseline_time'] = (\n",
    "    pd.to_datetime(df['intime']) +\n",
    "    pd.to_timedelta(df['baseline_hour'], unit='hr')\n",
    ")\n",
    "# Identify the stays corresponding to patients whose\n",
    "# deathtimes occured before the baseline time for the study.\n",
    "stays_to_omit = df.loc[ \n",
    "    df['baseline_time'] > pd.to_datetime(df['deathtime']),\n",
    "    'stay_id'\n",
    "].unique()\n",
    "\n",
    "df = df.loc[~df['stay_id'].isin(stays_to_omit)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, we'll filter the dataframe so only the first ICU stay for each patient is included."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Select subject_id and intime from each valid stay_id in cohort\n",
    "subjects = cohort.loc[\n",
    "    cohort['stay_id'].isin(df['stay_id']), ['subject_id', 'stay_id', 'intime']\n",
    "]\n",
    "# Sort by subject_id, then by intime to identify which is the first stay\n",
    "subjects.sort_values(['subject_id', 'intime'], ignore_index=True, inplace=True)\n",
    "# Get those first stay_ids\n",
    "first_stay_ids = subjects.groupby('subject_id')['stay_id'].first()\n",
    "# Filter df (and subjects) to only include those stays\n",
    "df = df[df['stay_id'].isin(first_stay_ids)]\n",
    "subjects = subjects[subjects['stay_id'].isin(first_stay_ids)]\n",
    "\n",
    "# Sort the df\n",
    "df.sort_values(['stay_id', 'hour'], ignore_index=True, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Identify variables whose NA proportion is above 0.10, so we can exclude them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "marital_status: 0.15031893611692573\n"
     ]
    }
   ],
   "source": [
    "cols_to_exclude = []\n",
    "for c in ALL_WEANING_PREDICTORS:\n",
    "    na_proportion = df[c].isna().mean()\n",
    "    if na_proportion >= 0.1:\n",
    "        cols_to_exclude.append(c)\n",
    "        print(f'{c}: {na_proportion}')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that we've identified which columns should be excluded from our analysis, we can create a new list of predictors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "weaning_predictors = [\n",
    "    c for c in ALL_WEANING_PREDICTORS if c not in cols_to_exclude\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Impute missing values (for numeric columns) with the median value."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "for c in weaning_predictors:\n",
    "    try:\n",
    "        df[c].fillna(df[c].median(), inplace=True)\n",
    "    # If a column is a string or object type, the median method will throw a TypeError\n",
    "    except TypeError:\n",
    "        pass"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we'll standardize our predictors."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "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>mean</th>\n",
       "      <th>std</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>last_peep_set</th>\n",
       "      <td>1.894669e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_tidal_volume_set</th>\n",
       "      <td>-2.701398e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_resp_rate</th>\n",
       "      <td>-3.831521e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_peak_insp_pressure</th>\n",
       "      <td>-1.344547e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_plateau_pressure</th>\n",
       "      <td>-1.963215e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_mean_airway_pressure</th>\n",
       "      <td>2.193458e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_minutes_vol</th>\n",
       "      <td>2.293640e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_pao2fio2ratio</th>\n",
       "      <td>3.485278e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_driving_pressure</th>\n",
       "      <td>-1.428911e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_po2</th>\n",
       "      <td>-5.769250e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_pco2</th>\n",
       "      <td>1.613457e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_aado2_calc</th>\n",
       "      <td>2.530913e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_ph</th>\n",
       "      <td>5.858711e-15</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_baseexcess</th>\n",
       "      <td>7.996981e-18</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_totalco2</th>\n",
       "      <td>5.448492e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_temperature</th>\n",
       "      <td>1.020221e-14</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_fio2_chartevents</th>\n",
       "      <td>1.986063e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_urine_output</th>\n",
       "      <td>-5.362811e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_lactate</th>\n",
       "      <td>1.572593e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_glucose</th>\n",
       "      <td>-3.954551e-19</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_gcs</th>\n",
       "      <td>1.794487e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_gcs_verbal</th>\n",
       "      <td>-3.154853e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_gcs_motor</th>\n",
       "      <td>3.093338e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_gcs_eyes</th>\n",
       "      <td>-2.602973e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_gcs_unable</th>\n",
       "      <td>1.971124e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_CREATININE</th>\n",
       "      <td>-3.047641e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_PLATELET</th>\n",
       "      <td>7.487283e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_PTT</th>\n",
       "      <td>2.871004e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_BUN</th>\n",
       "      <td>2.221579e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_WBC</th>\n",
       "      <td>-6.784252e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_sbp</th>\n",
       "      <td>-4.312218e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_dbp</th>\n",
       "      <td>-1.117820e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_mbp</th>\n",
       "      <td>3.265580e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_spo2</th>\n",
       "      <td>-4.924734e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_heart_rate</th>\n",
       "      <td>-3.100368e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_respiration_24hours</th>\n",
       "      <td>-6.204251e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_coagulation_24hours</th>\n",
       "      <td>-1.652124e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_liver_24hours</th>\n",
       "      <td>-2.478185e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_cardiovascular_24hours</th>\n",
       "      <td>3.620611e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_cns_24hours</th>\n",
       "      <td>-1.040486e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_renal_24hours</th>\n",
       "      <td>1.379699e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_amount</th>\n",
       "      <td>-7.096222e-18</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_weight</th>\n",
       "      <td>-1.021153e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_imputed_TV_standardized</th>\n",
       "      <td>-1.574526e-15</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_opioid</th>\n",
       "      <td>2.794549e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_benzo</th>\n",
       "      <td>6.784252e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_propofol</th>\n",
       "      <td>1.031698e-16</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>last_dex</th>\n",
       "      <td>-6.687585e-17</td>\n",
       "      <td>1.000002</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                      mean       std\n",
       "last_peep_set                 1.894669e-16  1.000002\n",
       "last_tidal_volume_set        -2.701398e-16  1.000002\n",
       "last_resp_rate               -3.831521e-17  1.000002\n",
       "last_peak_insp_pressure      -1.344547e-16  1.000002\n",
       "last_plateau_pressure        -1.963215e-16  1.000002\n",
       "last_mean_airway_pressure     2.193458e-16  1.000002\n",
       "last_minutes_vol              2.293640e-17  1.000002\n",
       "last_pao2fio2ratio            3.485278e-16  1.000002\n",
       "last_driving_pressure        -1.428911e-16  1.000002\n",
       "last_po2                     -5.769250e-17  1.000002\n",
       "last_pco2                     1.613457e-16  1.000002\n",
       "last_aado2_calc               2.530913e-17  1.000002\n",
       "last_ph                       5.858711e-15  1.000002\n",
       "last_baseexcess               7.996981e-18  1.000002\n",
       "last_totalco2                 5.448492e-17  1.000002\n",
       "last_temperature              1.020221e-14  1.000002\n",
       "last_fio2_chartevents         1.986063e-16  1.000002\n",
       "last_urine_output            -5.362811e-17  1.000002\n",
       "last_lactate                  1.572593e-16  1.000002\n",
       "last_glucose                 -3.954551e-19  1.000002\n",
       "last_gcs                      1.794487e-16  1.000002\n",
       "last_gcs_verbal              -3.154853e-17  1.000002\n",
       "last_gcs_motor                3.093338e-17  1.000002\n",
       "last_gcs_eyes                -2.602973e-16  1.000002\n",
       "last_gcs_unable               1.971124e-16  1.000002\n",
       "last_CREATININE              -3.047641e-16  1.000002\n",
       "last_PLATELET                 7.487283e-17  1.000002\n",
       "last_PTT                      2.871004e-16  1.000002\n",
       "last_BUN                      2.221579e-16  1.000002\n",
       "last_WBC                     -6.784252e-17  1.000002\n",
       "last_sbp                     -4.312218e-16  1.000002\n",
       "last_dbp                     -1.117820e-16  1.000002\n",
       "last_mbp                      3.265580e-16  1.000002\n",
       "last_spo2                    -4.924734e-16  1.000002\n",
       "last_heart_rate              -3.100368e-16  1.000002\n",
       "last_respiration_24hours     -6.204251e-17  1.000002\n",
       "last_coagulation_24hours     -1.652124e-17  1.000002\n",
       "last_liver_24hours           -2.478185e-17  1.000002\n",
       "last_cardiovascular_24hours   3.620611e-17  1.000002\n",
       "last_cns_24hours             -1.040486e-16  1.000002\n",
       "last_renal_24hours            1.379699e-17  1.000002\n",
       "last_amount                  -7.096222e-18  1.000002\n",
       "last_weight                  -1.021153e-16  1.000002\n",
       "last_imputed_TV_standardized -1.574526e-15  1.000002\n",
       "last_opioid                   2.794549e-17  1.000002\n",
       "last_benzo                    6.784252e-17  1.000002\n",
       "last_propofol                 1.031698e-16  1.000002\n",
       "last_dex                     -6.687585e-17  1.000002"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Get the numeric columns\n",
    "numeric = df[weaning_predictors].select_dtypes(exclude='object').columns\n",
    "# Standardize only the columns tracking the 'last' values of a variable\n",
    "to_standardize = [c for c in numeric if 'last' in c]\n",
    "df.loc[:, to_standardize] = StandardScaler().fit_transform(df[to_standardize])\n",
    "# Sanity check to make sure the DataFrame was actually standardized\n",
    "df[to_standardize].agg(['mean', 'std']).T"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For our final processing step, we'll remove extreme values from the covariates."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "for c in weaning_predictors:\n",
    "    if 'last' in c:\n",
    "        df[c] = np.maximum(-5, np.minimum(df[c], 5))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Define patient subgroups by risk\n",
    "We'll later be performing a subgroup analysis, so we divide up our patients based on risk indicators like P/F ratio and SOFA score."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_pats(cond: pd.Series) -> pd.Series:\n",
    "    \"\"\"Return the stay_ids for patients meeting the given condition.\"\"\"\n",
    "    rows = (df['hour_baseline'] == 1) & cond\n",
    "    return df.loc[rows, 'stay_id'].dropna().unique()\n",
    "\n",
    "low_pfratio_pats = get_pats(df['pao2fio2ratio'] <= 150)\n",
    "high_pfratio_pats = get_pats(df['pao2fio2ratio'] >= 300)\n",
    "\n",
    "low_sofa_pats = get_pats(df['sofa_24hours'] <= 5)\n",
    "high_sofa_pats = get_pats(df['sofa_24hours'] >= 12)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Next, load in pulmonary edema data and process it to define ARDS cohorts."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>subject_id</th>\n",
       "      <th>hadm_id</th>\n",
       "      <th>note_id</th>\n",
       "      <th>charttime</th>\n",
       "      <th>note_type</th>\n",
       "      <th>field_value</th>\n",
       "      <th>No Finding</th>\n",
       "      <th>Enlarged Cardiomediastinum</th>\n",
       "      <th>Cardiomegaly</th>\n",
       "      <th>Lung Lesion</th>\n",
       "      <th>Lung Opacity</th>\n",
       "      <th>Edema</th>\n",
       "      <th>Consolidation</th>\n",
       "      <th>Pneumonia</th>\n",
       "      <th>Atelectasis</th>\n",
       "      <th>Pneumothorax</th>\n",
       "      <th>Pleural Effusion</th>\n",
       "      <th>Pleural Other</th>\n",
       "      <th>Fracture</th>\n",
       "      <th>Support Devices</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>16977294</td>\n",
       "      <td>29291018.0</td>\n",
       "      <td>16977294-RR-17</td>\n",
       "      <td>2122-06-17 21:49:00</td>\n",
       "      <td>RR</td>\n",
       "      <td>C12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>12233384</td>\n",
       "      <td>22824404.0</td>\n",
       "      <td>12233384-RR-78</td>\n",
       "      <td>2170-10-08 12:25:00</td>\n",
       "      <td>RR</td>\n",
       "      <td>C11</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>-1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>11571683</td>\n",
       "      <td>22882793.0</td>\n",
       "      <td>11571683-RR-52</td>\n",
       "      <td>2138-07-05 15:02:00</td>\n",
       "      <td>RR</td>\n",
       "      <td>C11</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>17540015</td>\n",
       "      <td>21531298.0</td>\n",
       "      <td>17540015-RR-77</td>\n",
       "      <td>2189-10-14 04:45:00</td>\n",
       "      <td>RR</td>\n",
       "      <td>C12</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10150056</td>\n",
       "      <td>NaN</td>\n",
       "      <td>10150056-RR-40</td>\n",
       "      <td>2151-09-08 11:39:00</td>\n",
       "      <td>RR</td>\n",
       "      <td>C11</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   subject_id     hadm_id         note_id            charttime note_type  \\\n",
       "0    16977294  29291018.0  16977294-RR-17  2122-06-17 21:49:00        RR   \n",
       "1    12233384  22824404.0  12233384-RR-78  2170-10-08 12:25:00        RR   \n",
       "2    11571683  22882793.0  11571683-RR-52  2138-07-05 15:02:00        RR   \n",
       "3    17540015  21531298.0  17540015-RR-77  2189-10-14 04:45:00        RR   \n",
       "4    10150056         NaN  10150056-RR-40  2151-09-08 11:39:00        RR   \n",
       "\n",
       "  field_value  No Finding  Enlarged Cardiomediastinum  Cardiomegaly  \\\n",
       "0         C12         NaN                         1.0           NaN   \n",
       "1         C11         NaN                         NaN           1.0   \n",
       "2         C11         NaN                         NaN           0.0   \n",
       "3         C12         NaN                         NaN           1.0   \n",
       "4         C11         NaN                         NaN           1.0   \n",
       "\n",
       "   Lung Lesion  Lung Opacity  Edema  Consolidation  Pneumonia  Atelectasis  \\\n",
       "0          NaN           1.0   -1.0            1.0        1.0          NaN   \n",
       "1          NaN          -1.0    1.0            NaN        NaN          1.0   \n",
       "2          NaN           NaN    NaN            0.0        NaN          NaN   \n",
       "3          NaN           1.0    NaN            NaN        NaN          NaN   \n",
       "4          NaN           NaN    NaN            NaN        NaN          1.0   \n",
       "\n",
       "   Pneumothorax  Pleural Effusion  Pleural Other  Fracture  Support Devices  \n",
       "0           0.0               NaN            NaN       NaN              NaN  \n",
       "1           0.0               0.0            NaN       NaN              1.0  \n",
       "2           0.0               0.0            NaN       NaN              NaN  \n",
       "3           NaN               1.0            NaN       NaN              1.0  \n",
       "4           0.0               1.0            NaN       NaN              NaN  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pulmonary_edema = pd.read_csv('mimic_iv_cxr_reports_labeled_notext.csv')\n",
    "pulmonary_edema.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# keep only rows when pulmonary edema is 1 or -1,\n",
    "# and cardiomegaly is not 1\n",
    "correct_vals = (pulmonary_edema['Edema'].isin((1, -1))) & (pulmonary_edema['Cardiomegaly'] != 1)\n",
    "# also make sure neither edema nor cardiomegaly columns are na\n",
    "not_na = pulmonary_edema[['Cardiomegaly', 'Edema']].notna().all(axis=1)\n",
    "# filter by the above conditions\n",
    "pulmonary_edema = pulmonary_edema.loc[correct_vals & not_na]\n",
    "# and merge in the following columns from cohort\n",
    "pulmonary_edema = pulmonary_edema.merge(\n",
    "    cohort[['subject_id', 'hadm_id', 'stay_id', 'intime', 'outtime']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert various time columns to pandas datetime objects\n",
    "for col in ('charttime', 'intime', 'outtime'):\n",
    "    pulmonary_edema[col] = pd.to_datetime(pulmonary_edema[col])\n",
    "\n",
    "# create new columns tracking duration between charttime and intime, and outtime and intime\n",
    "pulmonary_edema['charthour'] = pulmonary_edema['charttime'] - pulmonary_edema['intime']\n",
    "pulmonary_edema['outhour'] = pulmonary_edema['outtime'] - pulmonary_edema['intime']\n",
    "\n",
    "# only include records where charthour < outhour\n",
    "pulmonary_edema = pulmonary_edema.loc[pulmonary_edema['charthour'] < pulmonary_edema['outhour']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# create new table tracking only essential values\n",
    "# from pulmonary_edema\n",
    "edema_df = pulmonary_edema.loc[:, ['stay_id', 'charthour', 'Edema']]\n",
    "# group by stay_id and edema, and take the minimum value for each group\n",
    "edema_df = edema_df.groupby(['stay_id', 'Edema']).min().reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# convert charthour to an array of ints\n",
    "# counting the total hours the timedelta represents\n",
    "edema_df['charthour'] = (edema_df['charthour']\n",
    "    .dt.ceil(freq='H')\n",
    "    .dt.total_seconds()\n",
    "    / 3600\n",
    ").astype(int)\n",
    "# then set all negative values to 1\n",
    "edema_df['hour'] = np.maximum(1, edema_df['charthour'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "# set all values for pulmonary_edema to 1\n",
    "edema_df['Edema'] = 1\n",
    "\n",
    "# then merge in the data into our main dataframe\n",
    "edema_df.rename({'Edema': 'pulmonary_edema'}, axis=1, inplace=True)\n",
    "df = df.merge(edema_df[['stay_id', 'hour', 'pulmonary_edema']], how='left')\n",
    "\n",
    "# set all na values to zero\n",
    "df['pulmonary_edema'].fillna(0, inplace=True)\n",
    "\n",
    "# then set the values at each intermediate time point to be\n",
    "# the cumulative number of timesteps where pulmonary_edema was true\n",
    "df['pulmonary_edema'] = df.groupby('stay_id')['pulmonary_edema'].cumsum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define another subset of patients: those with ards\n",
    "ards_criteria = (df['pao2fio2ratio'] <= 300) & (df['pulmonary_edema'] == 1)\n",
    "ards_pats = df.loc[ards_criteria, 'stay_id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# now that we have our pulmonary edema data,\n",
    "# we'll add it to our covariates list\n",
    "weaning_predictors.append('pulmonary_edema')\n",
    "\n",
    "# save the processed data\n",
    "df.to_csv('processed_data.csv')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Fit weaning predictor\n",
    "We want to estimate the propensity $\\pi_t(h_t)$ of a patient to be weaned off of mechanical ventilation at any time step $t$ given covariate and treatment history $h_t$. Here, we'll fit a logistic regression model to estimate this quantity."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "MODEL_PARAMS = {\n",
    "    'penalty': 'l2',\n",
    "    'max_iter': 2000,\n",
    "}\n",
    "# Dataframe checkpoint (loading in full processed dataset)\n",
    "df = pd.read_csv('processed_data.csv')\n",
    "\n",
    "X = df.loc[:, weaning_predictors]\n",
    "y = df['weaning']\n",
    "\n",
    "# Encode string/object columns as integers\n",
    "for col in X.select_dtypes(include='object'):\n",
    "    X[col] = LabelEncoder().fit_transform(X[col])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "weaning_predictor = LogisticRegression(**MODEL_PARAMS).fit(X, y)\n",
    "\n",
    "with open('models/weaning_predictor.pkl', 'wb') as f:\n",
    "    pickle.dump(weaning_predictor, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "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": [
    "# Model checkpoint\n",
    "with open('models/weaning_predictor.pkl', 'rb') as f:\n",
    "    weaning_predictor = pickle.load(f)\n",
    "\n",
    "# create a calibration curve\n",
    "df['p_hat'] = weaning_predictor.predict_proba(X)[:, 1]\n",
    "\n",
    "calibration_curve(\n",
    "    df, 'p_hat', 'weaning', 'Logistic Regression Calibration Curve', \n",
    "    filename='logistic_reg_calibration.png'\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Train a calibrated regressor since the above graph shows that the current predictor is poorly calibrated."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "calibrated_predictor = CalibratedClassifierCV(\n",
    "    base_estimator=LogisticRegression(**MODEL_PARAMS),\n",
    "    method='isotonic'\n",
    "    ).fit(X, y)\n",
    "\n",
    "with open('models/calibrated_predictor.pkl', 'wb') as f:\n",
    "    pickle.dump(calibrated_predictor, f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "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": [
    "# Model checkpoint\n",
    "with open('models/calibrated_predictor.pkl', 'rb') as f:\n",
    "    calibrated_predictor = pickle.load(f)\n",
    "\n",
    "df['propensity_score'] = calibrated_predictor.predict_proba(X)[:, 1]\n",
    "\n",
    "calibration_curve(\n",
    "    df, 'propensity_score', 'weaning', 'Isotonic Logistic Regression Calibration Curve',\n",
    "    filename='isoreg_calibration.png'\n",
    "    )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Effect estimation\n",
    "With our well-calibrated model for the probability of weaning, we can now estimate the expected value of $Y$ under an intervention multiplying liberation propensity by $\\delta$ as follows:\n",
    "\n",
    "$$\\hat{\\psi}_Y(\\delta) = \\frac{1}{n} \\sum \\left( \\prod_{t} \\frac{\\delta A_t + 1 - A_t}{\\delta \\hat{\\pi}(h_t) + 1 - \\hat{\\pi}(h_t)} \\right) Y,$$\n",
    "where $ \\hat{\\pi}(h_t) $ represents the estimated conditional probability of liberation at time $t$, which we obtained above from the calibrated logistic regression model.\n",
    "\n",
    "Breaking down this formula yields four major steps for each value of $\\delta$.\n",
    "\n",
    "1. Construct time-dependent weights $W_t = \\frac{\\delta A_t + 1 - A_t}{\\delta \\hat{\\pi}(h_t) + 1 - \\hat{\\pi}(h_t)}$ for each subject and timepoint\n",
    "2. Calculate the cumulative weight $\\widetilde{W}_t = \\prod_{i = 1}^t W_i$ for each subject across all timepoints\n",
    "3. Multiply each cumulative weight by the outcome $Y$\n",
    "4. Take the average of these products to estimate the mean counterfactual outcome $\\hat{\\psi}_Y(\\delta)$\n",
    "\n",
    "Since we have multiple outcomes of interest, we first calculate the cumulative weights and later multiply them by each outcome."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "# round to remove floating point errors\n",
    "DELTAS = np.around(np.arange(0.5, 2.1, 0.1), decimals=1)\n",
    "\n",
    "def cumulative_weights(data, truncated=True, propensity_col: str = 'propensity_score'):\n",
    "    \"\"\"\n",
    "    For each value of delta, calculate weights for each patient using the given\n",
    "    propensity scores.\n",
    "    \"\"\"\n",
    "    weights = {}\n",
    "    for delta in DELTAS:\n",
    "        A = data['weaning']\n",
    "        pi_hat = data[propensity_col]\n",
    "        weight_col = f'weights_d{delta}'\n",
    "        # construct time-dependent weights (W_t) for each subject at each timepoint\n",
    "        data.loc[:, weight_col] = (\n",
    "            ((delta * A) + 1 - A) /\n",
    "            ((delta * pi_hat) + 1 - pi_hat)\n",
    "        )\n",
    "        # calculate cumulative product of weights (\\tilde{W_t})\n",
    "        cumulative_products = group_by_stay(data, weight_col).prod()\n",
    "        # truncate cumulative weights at the 99th percentile\n",
    "        if truncated:\n",
    "            max_val = cumulative_products.quantile(q=0.99)\n",
    "            cumulative_products = np.minimum(cumulative_products, max_val)\n",
    "        # save the weights\n",
    "        weights[delta] = cumulative_products\n",
    "    return weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "# calculate the overall weights\n",
    "weights = cumulative_weights(df)\n",
    "\n",
    "# now breakdown by subgroup\n",
    "def weights_by_subgroup(stay_ids):\n",
    "    return {delta: col[stay_ids] for delta, col in weights.items()}\n",
    "\n",
    "weights_pf = weights_by_subgroup(low_pfratio_pats)\n",
    "weights_pf_high = weights_by_subgroup(high_pfratio_pats)\n",
    "weights_sofa = weights_by_subgroup(high_sofa_pats)\n",
    "weights_sofa_low = weights_by_subgroup(low_sofa_pats)\n",
    "weights_ards = weights_by_subgroup(ards_pats)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Time until weaning outcome\n",
    "First, examine the impact on time until weaning. The goal of this analysis is to make sure that our incremental interventions actually affect the counterfactual time until weaning for a given patient."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "# time until weaning outcome\n",
    "grouped = group_by_stay(df, 'weaning')\n",
    "df['time_until_wean'] = (grouped\n",
    "    # Apply the .max() and .size() functions to the weaning groups\n",
    "    # This determines 1. whether weaning occured\n",
    "    # and 2. the number of hours at which weaning began (if it did)\n",
    "    .agg(['max', 'size'])\n",
    "    # If weaning did occur, return the number of hours after baseline with it occured\n",
    "    # Otherwise, we set the value to 720 hours\n",
    "    .apply(lambda row: row['size'] if row['max'] else 720, axis=1)\n",
    "    # Repeat rows to ensure the size of the array matches what's in the dataframe\n",
    "    .repeat(grouped.size())\n",
    "    # Reset index so column can be inserted into the dataframe properly\n",
    "    .reset_index(drop=True)\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Define a helper function to calculate our estimates given a set of cumulative weights and a given outcome."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "def calculate_estimates(weights, outcome):\n",
    "    \"\"\"\n",
    "    Estimate the effect on the given outcome of intervening on weawning.\n",
    "    \"\"\"\n",
    "    if isinstance(outcome, pd.core.groupby.generic.SeriesGroupBy):\n",
    "        outcome = outcome.first()\n",
    "    results = {}\n",
    "    for delta in DELTAS:\n",
    "        # multiply by outcome\n",
    "        weighted_products = weights[delta].multiply(outcome)\n",
    "        # average over subjects and add to results\n",
    "        results[delta] = weighted_products.mean()\n",
    "    return results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0.5: 73.17804449216197,\n",
      " 0.6: 65.84771945407431,\n",
      " 0.7: 59.116308042511676,\n",
      " 0.8: 53.24077883791837,\n",
      " 0.9: 48.25067452639452,\n",
      " 1.0: 44.05771559262774,\n",
      " 1.1: 40.21565674731608,\n",
      " 1.2: 37.22467854882971,\n",
      " 1.3: 34.76077641841339,\n",
      " 1.4: 32.66538157361617,\n",
      " 1.5: 30.847460733613048,\n",
      " 1.6: 29.247992328081157,\n",
      " 1.7: 27.82550949893563,\n",
      " 1.8: 26.549351828072524,\n",
      " 1.9: 25.396068061869638,\n",
      " 2.0: 24.347280958565097}\n"
     ]
    }
   ],
   "source": [
    "wean_times_results = calculate_estimates(weights, group_by_stay(df, 'time_until_wean'))\n",
    "pprint(wean_times_results)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, break down the results by subgroup. First, we'll define some helper functions to calculate estimates for each subgroup and display our results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for convenience, name each subgroup\n",
    "# (the groups are constant so this won't cause problems)\n",
    "SUBGROUPS = (\n",
    "    {  \n",
    "        'SOFA >= 12': high_sofa_pats,\n",
    "        'SOFA <= 5': low_sofa_pats,\n",
    "    },\n",
    "    {\n",
    "        'PaO2/FiO2 Ratio <= 150': low_pfratio_pats,\n",
    "        'PaO2/FiO2 Ratio >= 300': high_pfratio_pats,\n",
    "    }\n",
    ")\n",
    "\n",
    "def get_subgroup_results(data: pd.DataFrame, outcome_col: str):\n",
    "    \"\"\"\n",
    "    Given a dataframe with the stay_id column and a specified outcome column,\n",
    "    return a dictionary of the results broken down by each subgroup.\n",
    "    \"\"\"\n",
    "    results = {}\n",
    "    groups = {**SUBGROUPS[0], **SUBGROUPS[1]}\n",
    "    for name, pats in groups.items():\n",
    "        subgroup_weights = weights_by_subgroup(pats)\n",
    "        subgroup_data = data.loc[data['stay_id'].isin(pats)]\n",
    "        outcome = group_by_stay(subgroup_data, outcome_col)\n",
    "        results[name] = calculate_estimates(subgroup_weights, outcome)\n",
    "    return results\n",
    "\n",
    "def display_results(title, ylabel, cohort_results, subgroup_results, filename=''):\n",
    "    fig, axs = plt.subplots(nrows=2, ncols=1, figsize=(10, 10))\n",
    "    for grps, ax in zip(SUBGROUPS, axs):\n",
    "        # plot full cohort data on each subplot\n",
    "        ax.scatter(cohort_results.keys(), cohort_results.values(), color='black', label='Full cohort')\n",
    "        # then plot sofa on ax[0] and pf on ax[1]\n",
    "        for name in grps:\n",
    "            ax.scatter(\n",
    "                subgroup_results[name].keys(),\n",
    "                subgroup_results[name].values(),\n",
    "                label=name\n",
    "                )\n",
    "        # now for cosmetics\n",
    "        ax.set_title(title)\n",
    "        ax.set_xlabel('Delta')\n",
    "        ax.set_ylabel(ylabel)\n",
    "        # add a vertical line at x = 1 to each plot\n",
    "        ax.axvline(1, color='grey', alpha=0.5)\n",
    "        # and a legend\n",
    "        ax.legend()\n",
    "    \n",
    "    if filename:\n",
    "        plt.savefig(f'images/{filename}.png', facecolor='white')\n",
    "    \n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "wean_times_subgroup_results = get_subgroup_results(df, 'time_until_wean')\n",
    "\n",
    "display_results(\n",
    "    'Counterfactual Time to First Attempt Under Incremental Interventions',\n",
    "    'Hours to First Liberation Attempt',\n",
    "    wean_times_results,\n",
    "    wean_times_subgroup_results,\n",
    "    'wean_times'\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Mortality outcome\n",
    "Next, calculate the effect on mortality."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{0.5: 0.33519454698033735,\n",
      " 0.6: 0.3373248962201278,\n",
      " 0.7: 0.33730234644058643,\n",
      " 0.8: 0.336488820414818,\n",
      " 0.9: 0.33554249734297764,\n",
      " 1.0: 0.33472353020314904,\n",
      " 1.1: 0.3337681044675111,\n",
      " 1.2: 0.33329447693110337,\n",
      " 1.3: 0.3330667012836953,\n",
      " 1.4: 0.33296851671825806,\n",
      " 1.5: 0.332935215521177,\n",
      " 1.6: 0.3329279216723524,\n",
      " 1.7: 0.33292241066739797,\n",
      " 1.8: 0.3329034404683114,\n",
      " 1.9: 0.33286156708116055,\n",
      " 2.0: 0.3327912110627911}\n"
     ]
    }
   ],
   "source": [
    "mortality_results = calculate_estimates(weights, group_by_stay(df, 'hospital_expire_flag'))\n",
    "mortality_subgroup_results = get_subgroup_results(df, 'hospital_expire_flag')\n",
    "\n",
    "pprint(mortality_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_results(\n",
    "    'Counterfactual Mortality Rates Under Incremental Interventions',\n",
    "    'In-Hospital Mortality Rate',\n",
    "    mortality_results,\n",
    "    mortality_subgroup_results,\n",
    "    'mortality'\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### ICU-free days outcome\n",
    "Basic structure:\n",
    "1. find baseline time\n",
    "2. determine how many of the 30 post-baseline days the patient was alive and not in the ICU\n",
    "\n",
    "\n",
    "Assumptions:\n",
    "- if the patient didn't die in a hospital stay, they didn't die\n",
    "- if they died in hospital, assume discharge time is time of death"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "\n",
    "# make sure dtypes agree\n",
    "df['baseline_time'] = pd.to_datetime(df['baseline_time'])\n",
    "\n",
    "for time_col in ('intime', 'outtime', 'deathtime'):\n",
    "    cohort[time_col] = pd.to_datetime(cohort[time_col])\n",
    "\n",
    "# Merge dataframe containing in and out time\n",
    "baseline_hour_df = cohort.loc[\n",
    "    :, ['stay_id', 'intime', 'outtime', 'subject_id']\n",
    "    ].merge(\n",
    "        df.loc[df['hour_baseline'] == 1, ['stay_id', 'baseline_hour', 'baseline_time']]\n",
    "    )\n",
    "\n",
    "icu_duration_df = cohort.merge(\n",
    "    baseline_hour_df.loc[:, ['subject_id', 'baseline_time']].drop_duplicates()\n",
    "    )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [],
   "source": [
    "icu_duration_df['max_time'] = icu_duration_df['baseline_time'] + pd.Timedelta(30, unit=\"days\")\n",
    "# remove ones where outtime <= baseline_time\n",
    "icu_duration_df = icu_duration_df.loc[icu_duration_df['outtime'] > icu_duration_df['baseline_time']]\n",
    "# remove rows with max_time <= intime\n",
    "icu_duration_df = icu_duration_df.loc[icu_duration_df['max_time'] > icu_duration_df['intime']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate duration of each stay\n",
    "icu_duration_df = icu_duration_df.sort_values('subject_id')\n",
    "stay_start = np.maximum(icu_duration_df['intime'], icu_duration_df['baseline_time'])\n",
    "stay_end = np.minimum(icu_duration_df['outtime'], icu_duration_df['max_time'])\n",
    "icu_duration_df['icu_stay_durations'] = stay_end - stay_start"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "icu_duration_df['death_day'] = icu_duration_df['deathtime'] - icu_duration_df['baseline_time']\n",
    "# fill all entries with deathtime before baseline_time with 0\n",
    "icu_duration_df.loc[icu_duration_df['death_day'] < pd.to_timedelta(0), 'death_day'] = pd.to_timedelta(0)\n",
    "# fill all entries with NaN deathtime with a dummy value\n",
    "icu_duration_df['death_day'].fillna(pd.Timedelta(100000, unit=\"days\"), inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "max_val = pd.Timedelta(30, \"days\")\n",
    "icu_duration_df['max'] = icu_duration_df['death_day'].where(\n",
    "    icu_duration_df['death_day'] <= max_val, max_val\n",
    "    )\n",
    "max_df = icu_duration_df[['subject_id', 'max']].drop_duplicates()\n",
    "max_df = max_df.groupby('subject_id', as_index=False).min()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "duration_df = icu_duration_df.groupby('subject_id', as_index=False)['icu_stay_durations'].sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "icu_outcome_df = max_df.merge(duration_df)\n",
    "icu_outcome_df.rename(columns = {'icu_stay_durations' : 'icu_duration'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [],
   "source": [
    "# construct outcome column (take ceil of the time difference to the nearest day)\n",
    "icu_outcome_df['outcome'] = (icu_outcome_df['max'] - icu_outcome_df['icu_duration']).dt.ceil('d')\n",
    "# and convert to an int\n",
    "icu_outcome_df['outcome'] = icu_outcome_df['outcome'].dt.days\n",
    "# merge with subjects to get stay id\n",
    "icu_outcome_df = icu_outcome_df.merge(subjects[['subject_id', 'stay_id']])\n",
    "icu_outcome_df.sort_values('stay_id', inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Calculate estimates!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "icu_free_days_results = calculate_estimates(weights, group_by_stay(icu_outcome_df, 'outcome'))\n",
    "icu_free_days_subgroup_results = get_subgroup_results(icu_outcome_df, 'outcome')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "display_results(\n",
    "    'Counterfactual ICU Free Days Under Incremental Interventions',\n",
    "    'ICU Free Days',\n",
    "    icu_free_days_results,\n",
    "    icu_free_days_subgroup_results,\n",
    "    filename='icufreedays'\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ventilator-free Days Outcome\n",
    "Similar to ICU-free days, we'll determine how many days in the first month post-baseline that the patient spends alive and off the ventilator."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load in our data\n",
    "vent = pd.read_csv('ventilator.csv')\n",
    "\n",
    "vent_duration_df = vent.merge(baseline_hour_df[['stay_id', 'subject_id', 'baseline_time']])\n",
    "vent_duration_df = vent_duration_df.loc[vent_duration_df['ventilation_status'].isin((\"InvasiveVent\", \"Trach\"))]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now create the `startday`, `endday`, and `death_day` columns which we'll use to create our outcome."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 144,
   "metadata": {},
   "outputs": [],
   "source": [
    "vent_duration_df['max_time'] = vent_duration_df['baseline_time'] + pd.Timedelta(30, \"days\")\n",
    "\n",
    "# convert endtime + max_time + starttime to datetime columns\n",
    "for time_col in ('starttime', 'endtime'):\n",
    "    vent_duration_df[time_col] = pd.to_datetime(vent_duration_df[time_col])\n",
    "\n",
    "# remove rows with endtime <= baseline_time\n",
    "vent_duration_df = vent_duration_df[vent_duration_df['endtime'] > vent_duration_df['baseline_time']]\n",
    "vent_duration_df = vent_duration_df[vent_duration_df['max_time'] > vent_duration_df['starttime']]\n",
    "\n",
    "# the ceil function rounds up to the nearest day, then use the dt.days accessor to convert to an integer\n",
    "vent_duration_df['startday'] = (vent_duration_df['starttime'] - vent_duration_df['baseline_time']).dt.ceil('D').dt.days\n",
    "vent_duration_df['endday'] = (vent_duration_df['endtime'] - vent_duration_df['baseline_time']).dt.ceil('D').dt.days\n",
    "\n",
    "# min startday at 1\n",
    "vent_duration_df.loc[vent_duration_df['startday'] < 1, 'startday'] = 1\n",
    "# cap endday at 30\n",
    "vent_duration_df.loc[vent_duration_df['endday'] > 30, 'endday'] = 30\n",
    "\n",
    "# add 1 to endday inside the arange function to include the endpoint\n",
    "vent_duration_df['vent_days'] = vent_duration_df.apply(lambda r: np.arange(r['startday'], r['endday'] + 1), axis=1)\n",
    "\n",
    "death_cohort = cohort[~cohort['deathtime'].isna()]\n",
    "\n",
    "baseline_hour_df = baseline_hour_df.merge(death_cohort[['subject_id', 'deathtime']].drop_duplicates(), how='left')\n",
    "baseline_hour_df['deathtime'] = pd.to_datetime(baseline_hour_df['deathtime'])\n",
    "\n",
    "baseline_hour_df['death_day'] = (baseline_hour_df['deathtime'] - baseline_hour_df['baseline_time']).dt.ceil('D').dt.days\n",
    "# min death_day at 1\n",
    "baseline_hour_df.loc[baseline_hour_df['death_day'] < 1, 'death_day'] = 1\n",
    "baseline_hour_df['death_day'].fillna(10000000, inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Use those columns defined above to track days patients spend ventilated (`vent_days`) or dead (`death_days`), and combine these to create our outcome."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 145,
   "metadata": {},
   "outputs": [],
   "source": [
    "patients = vent_duration_df['stay_id'].unique()\n",
    "\n",
    "# create a blank dataframe indexed by stay_id\n",
    "vent_days_df = pd.DataFrame(index=patients)\n",
    "vent_days_df.index.rename('stay_id', inplace=True)\n",
    "# create the outcome\n",
    "vent_days_df['vent_days'] = group_by_stay(vent_duration_df, 'vent_days').apply(lambda g: np.concatenate([entry for entry in g]))\n",
    "# reset the stay_id index to a column so it can be merged upon\n",
    "vent_days_df.reset_index(inplace=True)\n",
    "# merge it in\n",
    "baseline_hour_df = baseline_hour_df.merge(vent_days_df, on='stay_id', how='left')\n",
    "# replace NaN entries with an empty list\n",
    "baseline_hour_df['vent_days'] = baseline_hour_df['vent_days'].fillna('').apply(list) # first fill with '' bc lists cant be passed to fillna\n",
    "\n",
    "baseline_hour_df['death_days'] = baseline_hour_df.apply(lambda r: np.arange(r['death_day'], 31), axis=1)\n",
    "\n",
    "def get_vent_free_days(row):\n",
    "    \"\"\"Get the ventilator free days outcome for the given row.\"\"\"\n",
    "    days = np.concatenate((row['vent_days'], row['death_days']))\n",
    "    # remove duplicates\n",
    "    days = list(set(days))\n",
    "    # get the outcome\n",
    "    return 30 - len(days)\n",
    "\n",
    "baseline_hour_df['vent_free_days'] = baseline_hour_df.apply(get_vent_free_days, axis=1)\n",
    "\n",
    "baseline_hour_df.sort_values('stay_id', inplace=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Calculate estimates for ventilator-free days."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 152,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "vent_free_days_results = calculate_estimates(weights, group_by_stay(baseline_hour_df, 'vent_free_days'))\n",
    "vent_free_days_subgroup_results = get_subgroup_results(baseline_hour_df, 'vent_free_days')\n",
    "\n",
    "display_results(\n",
    "    'Counterfactual Ventilator Free Days Under Incremental Interventions',\n",
    "    'Ventilator Free Days',\n",
    "    vent_free_days_results,\n",
    "    vent_free_days_subgroup_results,\n",
    "    filename='ventfreedays'\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Uncertainty Estimation\n",
    "TODO: Bootstrap on outcomes to determine variability."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "N_BOOTSTRAPS = 100\n",
    "stays = df['stay_id'].unique()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "N_BOOTSTRAPS = 100\n",
    "outcome_estimates = {d: [] for d in DELTAS}\n",
    "\n",
    "stays = df['stay_id'].unique()\n",
    "\n",
    "def get_bootstrap_sample():\n",
    "    \"\"\"Randomly sample from the dataframe, keeping the number of stays the same.\"\"\"\n",
    "    sampled_stays = np.random.choice(stays, size=stays.shape, replace=True)\n",
    "    # Determine how many times each stay should be repeated (since we sample w/replacement)\n",
    "    num_repeats = {s: 0 for s in sampled_stays}\n",
    "    for s in sampled_stays:\n",
    "        num_repeats[s] += 1\n",
    "    # Choose only the selected stays\n",
    "    curr_data = df.loc[df['stay_id'].isin(sampled_stays)]\n",
    "    indices = []\n",
    "    for stay, idxs in group_by_stay(curr_data).groups.items():\n",
    "        indices += (num_repeats[stay] * list(idxs))\n",
    "    return curr_data.reindex(indices)\n",
    "\n",
    "for i in range(N_BOOTSTRAPS):\n",
    "    r = calculate_estimates(\n",
    "        DELTAS,\n",
    "        data=get_bootstrap_sample(),\n",
    "    )\n",
    "    for delta, estimate in r.items():\n",
    "        outcome_estimates[delta].append(estimate)\n",
    "\n",
    "outcome_estimates = pd.DataFrame(outcome_estimates)\n",
    "# outcome_estimates.to_csv('outcome_estimates.csv')\n",
    "outcome_estimates.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "to_plot = outcome_estimates.iloc[:, 2:17]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "bounds = outcome_estimates.quantile(0.025), outcome_estimates.quantile(0.975)\n",
    "# bounds = outcome_estimates.min(), outcome_estimates.max()\n",
    "plt.fill_between(delta_vals, bounds[0], bounds[1], alpha=0.5)\n",
    "plt.plot(mortality_results.keys(), mortality_results.values())\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "8736e52e49f3375c6a349f253dc148135d5c4698b23d6afe3435df37d8fb6c09"
  },
  "kernelspec": {
   "display_name": "Python 3.9.0 ('venv': venv)",
   "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": "3.9.0"
  },
  "metadata": {
   "interpreter": {
    "hash": "974242ba8bd4aee876d4cc6b2b4fd6d90080efacd3e889d4fd5fe2f92b3a2fbb"
   }
  },
  "orig_nbformat": 2
 },
 "nbformat": 4,
 "nbformat_minor": 2
}