2.5_baseline_results.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Libraries\n",
    "import numpy as np\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "import math\n",
    "import time\n",
    "import random\n",
    "import sklearn\n",
    "from torch.utils.data import Dataset, DataLoader\n",
    "import pandas as pd\n",
    "import torch\n",
    "from torch import nn\n",
    "from torch import optim\n",
    "from tqdm import tqdm\n",
    "import torch.nn.functional as F\n",
    "\n",
    "from IPython.core.interactiveshell import InteractiveShell\n",
    "InteractiveShell.ast_node_interactivity = \"all\"\n",
    "\n",
    "# Remove printing error\n",
    "pd.options.mode.chained_assignment = None"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Data\n",
    "mimic_data = pd.read_csv('mimic_data.csv')\n",
    "\n",
    "\n",
    "# Split into 'preprocessing dataset' and a 'train/vaid/test dataset'\n",
    "stays = mimic_data['stay_id'].unique()\n",
    "\n",
    "random.Random(0).shuffle(stays)\n",
    "model_data2 = mimic_data.set_index(\"stay_id\").loc[stays].reset_index()\n",
    "n = round(0.5 * len(stays))\n",
    "preprocessing_stays = stays[:n]\n",
    "hold_out_stays = stays[n:]\n",
    "preprocessing_data = model_data2[model_data2['stay_id'].isin(preprocessing_stays)]\n",
    "hold_out_data = model_data2[model_data2['stay_id'].isin(hold_out_stays)]\n",
    "\n",
    "eicu_data = pd.read_csv('eicu_data.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "preprocessing_data = preprocessing_data.iloc[:,:6]\n",
    "hold_out_data = hold_out_data.iloc[:,:6]\n",
    "eicu_data = eicu_data.iloc[:,:6]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# mimic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create list\n",
    "hold_out_data_list = hold_out_data.stay_id.unique().tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load filtered chartevents data\n",
    "iter_csv = pd.read_csv(r\"mimic-iv-2.0/icu/chartevents.csv\", iterator=True, chunksize=10000000)\n",
    "chartevents = pd.concat([chunk[chunk['stay_id'].isin(hold_out_data_list)] for chunk in iter_csv])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Find relevant codes\n",
    "d_items = pd.read_csv(r\"mimic-iv-2.0/icu/d_items.csv\")\n",
    "\n",
    "pattern = 'Heart.Rate|temperature|febrile|WBC|C.Reactive.Protein|CRP|blood.pressure|respiratory.rate|Conscious|Richmond-RAS Scale|GCS -' # Added Richmond-RAS Scale and GCS -\n",
    "\n",
    "filtered_d_items = d_items[d_items['label'].str.contains(pattern, case=False, na=False)]\n",
    "\n",
    "filtered_d_items_list = filtered_d_items['itemid'].unique().tolist()\n",
    "\n",
    "# Filter for relevant codes\n",
    "filtered_chartevents = chartevents[chartevents['itemid'].isin(filtered_d_items_list)]\n",
    "\n",
    "filtered_chartevents = filtered_chartevents.merge(filtered_d_items[['itemid', 'label']], how='left', on=['itemid'])\n",
    "\n",
    "# Convert Farenhight to celcius \n",
    "filtered_chartevents.loc[filtered_chartevents['label'].str.contains('Temperature Fahrenheit'), 'valuenum'] = (filtered_chartevents.loc[filtered_chartevents['label'].str.contains('Temperature Fahrenheit'), 'valuenum']-32)*(5/9)\n",
    "\n",
    "# Drop PAR-Consciousness, Temperature Site, Changes in Temperature, Blood Pressure Alarm\n",
    "drop_pattern = 'PAR-Consciousness|Temperature Site|Changes in Temperature|Blood Pressure Alarm'\n",
    "filtered_chartevents2 = filtered_chartevents[~filtered_chartevents['label'].str.contains(drop_pattern, case=False, na=False)]\n",
    "\n",
    "string_list = [\n",
    " 'Heart rate Alarm - High',\n",
    " 'Heart Rate Alarm - Low',\n",
    " 'Heart Rate',\n",
    " 'Respiratory Rate .spontaneous',\n",
    " 'Respiratory Rate .Set',\n",
    " 'Respiratory Rate',\n",
    " 'Skin Temperature',\n",
    " 'Temperature',\n",
    " 'Blood Pressure systolic',\n",
    " 'Blood Pressure diastolic',\n",
    " 'Blood Pressure mean',\n",
    " 'WBC',\n",
    " 'Level of Consciousness',\n",
    " 'CRP',\n",
    " 'Febrile last 24 hours',\n",
    " 'GCS - Eye Opening',\n",
    " 'GCS - Verbal Response',\n",
    " 'GCS - Motor Response',\n",
    " 'Richmond-RAS Scale']\n",
    "\n",
    "pd.options.mode.chained_assignment = None\n",
    "\n",
    "x = 0\n",
    "for string in string_list:\n",
    "    x += 1\n",
    "    sub_df = filtered_chartevents2[filtered_chartevents2['label'].str.contains(string, case=False, na=False)]\n",
    "    sub_df['final_label'] = string.lower() # use filter string as final_label \n",
    "    if x == 1:\n",
    "        filtered_chartevents3 = sub_df\n",
    "    else:\n",
    "        filtered_chartevents3 = pd.concat([filtered_chartevents3, sub_df])\n",
    "\n",
    "# Need to drop rows labeled skin temp but actually skin temperature (duplicates)\n",
    "filtered_chartevents3 = filtered_chartevents3.loc[~((filtered_chartevents3['final_label'] == 'temperature') & (filtered_chartevents3['valuenum'].isna())),:]\n",
    "\n",
    "# O2 + CO2\n",
    "d_items_list = [220235, 220277, 223835]\n",
    "o2_chartevents = chartevents[chartevents['itemid'].isin(d_items_list)]\n",
    "o2_chartevents = o2_chartevents.merge(d_items[['itemid', 'label']], how='left', on=['itemid'])\n",
    "o2_chartevents['final_label'] = o2_chartevents['label'].str.lower()\n",
    "filtered_chartevents4 = pd.concat([filtered_chartevents3, o2_chartevents], axis=0)\n",
    "# Select relevant labels \n",
    "final_label_list = ['heart rate', 'respiratory rate', 'blood pressure mean', 'blood pressure systolic', 'blood pressure diastolic', 'temperature', 'wbc', 'o2 saturation pulseoxymetry', 'inspired o2 fraction', 'arterial co2 pressure']\n",
    "selected_catch22 = filtered_chartevents4[filtered_chartevents4['final_label'].isin(final_label_list)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "heart rate                     413115\n",
       "respiratory rate               404649\n",
       "blood pressure mean            345661\n",
       "blood pressure systolic        345425\n",
       "blood pressure diastolic       345358\n",
       "o2 saturation pulseoxymetry    341325\n",
       "temperature                    110461\n",
       "inspired o2 fraction            39884\n",
       "wbc                             21647\n",
       "arterial co2 pressure           18801\n",
       "Name: final_label, dtype: int64"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "selected_catch22.final_label.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 127,
   "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>stay_id</th>\n",
       "      <th>charttime</th>\n",
       "      <th>storetime</th>\n",
       "      <th>itemid</th>\n",
       "      <th>value</th>\n",
       "      <th>valuenum</th>\n",
       "      <th>valueuom</th>\n",
       "      <th>warning</th>\n",
       "      <th>label</th>\n",
       "      <th>final_label</th>\n",
       "      <th>date</th>\n",
       "      <th>hour</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-04 13:45:00</td>\n",
       "      <td>2129-08-04 14:15:00</td>\n",
       "      <td>220045</td>\n",
       "      <td>68</td>\n",
       "      <td>68.0</td>\n",
       "      <td>bpm</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Heart Rate</td>\n",
       "      <td>heart rate</td>\n",
       "      <td>2129-08-04</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-04 13:45:00</td>\n",
       "      <td>2129-08-04 14:16:00</td>\n",
       "      <td>220046</td>\n",
       "      <td>120</td>\n",
       "      <td>120.0</td>\n",
       "      <td>bpm</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Heart rate Alarm - High</td>\n",
       "      <td>heart rate</td>\n",
       "      <td>2129-08-04</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-04 13:45:00</td>\n",
       "      <td>2129-08-04 14:16:00</td>\n",
       "      <td>220047</td>\n",
       "      <td>50</td>\n",
       "      <td>50.0</td>\n",
       "      <td>bpm</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Heart Rate Alarm - Low</td>\n",
       "      <td>heart rate</td>\n",
       "      <td>2129-08-04</td>\n",
       "      <td>13</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-04 14:00:00</td>\n",
       "      <td>2129-08-04 14:18:00</td>\n",
       "      <td>220045</td>\n",
       "      <td>83</td>\n",
       "      <td>83.0</td>\n",
       "      <td>bpm</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Heart Rate</td>\n",
       "      <td>heart rate</td>\n",
       "      <td>2129-08-04</td>\n",
       "      <td>14</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-04 14:15:00</td>\n",
       "      <td>2129-08-04 14:15:00</td>\n",
       "      <td>220045</td>\n",
       "      <td>71</td>\n",
       "      <td>71.0</td>\n",
       "      <td>bpm</td>\n",
       "      <td>0.0</td>\n",
       "      <td>Heart Rate</td>\n",
       "      <td>heart rate</td>\n",
       "      <td>2129-08-04</td>\n",
       "      <td>14</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400005</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-25 15:00:00</td>\n",
       "      <td>2172-08-25 15:07:00</td>\n",
       "      <td>220277</td>\n",
       "      <td>99</td>\n",
       "      <td>99.0</td>\n",
       "      <td>%</td>\n",
       "      <td>0.0</td>\n",
       "      <td>O2 saturation pulseoxymetry</td>\n",
       "      <td>o2 saturation pulseoxymetry</td>\n",
       "      <td>2172-08-25</td>\n",
       "      <td>15</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400006</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-25 16:03:00</td>\n",
       "      <td>2172-08-25 16:34:00</td>\n",
       "      <td>220277</td>\n",
       "      <td>97</td>\n",
       "      <td>97.0</td>\n",
       "      <td>%</td>\n",
       "      <td>0.0</td>\n",
       "      <td>O2 saturation pulseoxymetry</td>\n",
       "      <td>o2 saturation pulseoxymetry</td>\n",
       "      <td>2172-08-25</td>\n",
       "      <td>16</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400007</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-25 17:00:00</td>\n",
       "      <td>2172-08-25 17:06:00</td>\n",
       "      <td>220277</td>\n",
       "      <td>100</td>\n",
       "      <td>100.0</td>\n",
       "      <td>%</td>\n",
       "      <td>0.0</td>\n",
       "      <td>O2 saturation pulseoxymetry</td>\n",
       "      <td>o2 saturation pulseoxymetry</td>\n",
       "      <td>2172-08-25</td>\n",
       "      <td>17</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400008</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-25 18:00:00</td>\n",
       "      <td>2172-08-25 19:13:00</td>\n",
       "      <td>220277</td>\n",
       "      <td>99</td>\n",
       "      <td>99.0</td>\n",
       "      <td>%</td>\n",
       "      <td>0.0</td>\n",
       "      <td>O2 saturation pulseoxymetry</td>\n",
       "      <td>o2 saturation pulseoxymetry</td>\n",
       "      <td>2172-08-25</td>\n",
       "      <td>18</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>400009</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-25 19:00:00</td>\n",
       "      <td>2172-08-25 19:13:00</td>\n",
       "      <td>220277</td>\n",
       "      <td>99</td>\n",
       "      <td>99.0</td>\n",
       "      <td>%</td>\n",
       "      <td>0.0</td>\n",
       "      <td>O2 saturation pulseoxymetry</td>\n",
       "      <td>o2 saturation pulseoxymetry</td>\n",
       "      <td>2172-08-25</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2386326 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        subject_id   hadm_id   stay_id            charttime  \\\n",
       "0         10002155  23822395  33685454  2129-08-04 13:45:00   \n",
       "2         10002155  23822395  33685454  2129-08-04 13:45:00   \n",
       "3         10002155  23822395  33685454  2129-08-04 13:45:00   \n",
       "11        10002155  23822395  33685454  2129-08-04 14:00:00   \n",
       "18        10002155  23822395  33685454  2129-08-04 14:15:00   \n",
       "...            ...       ...       ...                  ...   \n",
       "400005    19995091  27602139  32457489  2172-08-25 15:00:00   \n",
       "400006    19995091  27602139  32457489  2172-08-25 16:03:00   \n",
       "400007    19995091  27602139  32457489  2172-08-25 17:00:00   \n",
       "400008    19995091  27602139  32457489  2172-08-25 18:00:00   \n",
       "400009    19995091  27602139  32457489  2172-08-25 19:00:00   \n",
       "\n",
       "                  storetime  itemid value  valuenum valueuom  warning  \\\n",
       "0       2129-08-04 14:15:00  220045    68      68.0      bpm      0.0   \n",
       "2       2129-08-04 14:16:00  220046   120     120.0      bpm      0.0   \n",
       "3       2129-08-04 14:16:00  220047    50      50.0      bpm      0.0   \n",
       "11      2129-08-04 14:18:00  220045    83      83.0      bpm      0.0   \n",
       "18      2129-08-04 14:15:00  220045    71      71.0      bpm      0.0   \n",
       "...                     ...     ...   ...       ...      ...      ...   \n",
       "400005  2172-08-25 15:07:00  220277    99      99.0        %      0.0   \n",
       "400006  2172-08-25 16:34:00  220277    97      97.0        %      0.0   \n",
       "400007  2172-08-25 17:06:00  220277   100     100.0        %      0.0   \n",
       "400008  2172-08-25 19:13:00  220277    99      99.0        %      0.0   \n",
       "400009  2172-08-25 19:13:00  220277    99      99.0        %      0.0   \n",
       "\n",
       "                              label                  final_label        date  \\\n",
       "0                        Heart Rate                   heart rate  2129-08-04   \n",
       "2           Heart rate Alarm - High                   heart rate  2129-08-04   \n",
       "3            Heart Rate Alarm - Low                   heart rate  2129-08-04   \n",
       "11                       Heart Rate                   heart rate  2129-08-04   \n",
       "18                       Heart Rate                   heart rate  2129-08-04   \n",
       "...                             ...                          ...         ...   \n",
       "400005  O2 saturation pulseoxymetry  o2 saturation pulseoxymetry  2172-08-25   \n",
       "400006  O2 saturation pulseoxymetry  o2 saturation pulseoxymetry  2172-08-25   \n",
       "400007  O2 saturation pulseoxymetry  o2 saturation pulseoxymetry  2172-08-25   \n",
       "400008  O2 saturation pulseoxymetry  o2 saturation pulseoxymetry  2172-08-25   \n",
       "400009  O2 saturation pulseoxymetry  o2 saturation pulseoxymetry  2172-08-25   \n",
       "\n",
       "        hour  \n",
       "0         13  \n",
       "2         13  \n",
       "3         13  \n",
       "11        14  \n",
       "18        14  \n",
       "...      ...  \n",
       "400005    15  \n",
       "400006    16  \n",
       "400007    17  \n",
       "400008    18  \n",
       "400009    19  \n",
       "\n",
       "[2386326 rows x 14 columns]"
      ]
     },
     "execution_count": 127,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "selected_catch22"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Min Max Temp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Filter\n",
    "hold_out_data2 = selected_catch22[(selected_catch22['final_label'] == 'temperature')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert to str or does not work\n",
    "hold_out_data.date = hold_out_data.date.astype(str)\n",
    "#hold_out_data2.date = hold_out_data2.date.astype(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Merge\n",
    "hold_out_data2 = pd.merge(hold_out_data, hold_out_data2[['stay_id', 'charttime', 'valuenum']])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Datetime\n",
    "import datetime\n",
    "hold_out_data2['charttime'] =  pd.to_datetime(hold_out_data2['charttime']).dt.date + datetime.timedelta(days=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert to str\n",
    "hold_out_data2['charttime'] = hold_out_data2['charttime'].astype(str)\n",
    "hold_out_data2['date'] = hold_out_data2['date'].astype(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Filter\n",
    "hold_out_data3 = hold_out_data2.loc[hold_out_data2['date'] == hold_out_data2['charttime']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0    61996\n",
       "1    17943\n",
       "Name: po_flag, dtype: int64"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hold_out_data3.po_flag.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Index(['stay_id', 'date', 'iv_flag', 'first_po_flag', 'po_flag',\n",
       "       'iv_treatment_length', 'charttime', 'valuenum'],\n",
       "      dtype='object')"
      ]
     },
     "execution_count": 134,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hold_out_data3.columns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def summary_function(df):\n",
    "    return df['valuenum'].agg(['min','max'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get max and min\n",
    "hold_out_data4 = hold_out_data3.groupby(['stay_id', 'date'], as_index=False).apply(summary_function)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Merge\n",
    "hold_out_data5 = pd.merge(hold_out_data3[['stay_id', 'date', 'iv_flag', 'first_po_flag', 'po_flag', 'iv_treatment_length']], hold_out_data4)\n",
    "hold_out_data5.drop_duplicates(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "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>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>iv_flag</th>\n",
       "      <th>first_po_flag</th>\n",
       "      <th>po_flag</th>\n",
       "      <th>iv_treatment_length</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-12</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>35.000000</td>\n",
       "      <td>37.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-13</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>36.888889</td>\n",
       "      <td>37.600000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-21</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>37.055556</td>\n",
       "      <td>37.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-22</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>36.944444</td>\n",
       "      <td>39.111111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>32889989</td>\n",
       "      <td>2126-06-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>35.600000</td>\n",
       "      <td>36.800000</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79912</th>\n",
       "      <td>30772812</td>\n",
       "      <td>2130-06-08</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>35.555556</td>\n",
       "      <td>37.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79922</th>\n",
       "      <td>37914144</td>\n",
       "      <td>2136-05-23</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>36.555556</td>\n",
       "      <td>36.888889</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79925</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>36.888889</td>\n",
       "      <td>38.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79927</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-19</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>36.555556</td>\n",
       "      <td>37.944444</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79934</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-20</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>36.611111</td>\n",
       "      <td>36.888889</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10936 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        stay_id        date  iv_flag  first_po_flag  po_flag  \\\n",
       "0      31658148  2154-01-12      1.0            1.0        0   \n",
       "18     31658148  2154-01-13      1.0            1.0        0   \n",
       "31     33816676  2117-08-21      1.0            1.0        0   \n",
       "36     33816676  2117-08-22      1.0            1.0        0   \n",
       "44     32889989  2126-06-04      1.0            1.0        0   \n",
       "...         ...         ...      ...            ...      ...   \n",
       "79912  30772812  2130-06-08      1.0            1.0        0   \n",
       "79922  37914144  2136-05-23      1.0            1.0        0   \n",
       "79925  37246494  2127-07-18      1.0            1.0        0   \n",
       "79927  37246494  2127-07-19      1.0            1.0        0   \n",
       "79934  37246494  2127-07-20      1.0            1.0        0   \n",
       "\n",
       "       iv_treatment_length        min        max  \n",
       "0                        1  35.000000  37.600000  \n",
       "18                       2  36.888889  37.600000  \n",
       "31                       0  37.055556  37.666667  \n",
       "36                       1  36.944444  39.111111  \n",
       "44                       1  35.600000  36.800000  \n",
       "...                    ...        ...        ...  \n",
       "79912                    1  35.555556  37.500000  \n",
       "79922                    1  36.555556  36.888889  \n",
       "79925                    1  36.888889  38.333333  \n",
       "79927                    2  36.555556  37.944444  \n",
       "79934                    3  36.611111  36.888889  \n",
       "\n",
       "[10936 rows x 8 columns]"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hold_out_data5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Pivot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create date column \n",
    "selected_catch22['date'] =  pd.to_datetime(selected_catch22['charttime']).dt.date\n",
    "# Create hour column\n",
    "selected_catch22['hour'] =  pd.to_datetime(selected_catch22['charttime']).dt.hour"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "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>final_label</th>\n",
       "      <th>subject_id</th>\n",
       "      <th>hadm_id</th>\n",
       "      <th>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>arterial_co2_pressure</th>\n",
       "      <th>blood_pressure_diastolic</th>\n",
       "      <th>blood_pressure_mean</th>\n",
       "      <th>blood_pressure_systolic</th>\n",
       "      <th>heart_rate</th>\n",
       "      <th>inspired_o2_fraction</th>\n",
       "      <th>o2_saturation_pulseoxymetry</th>\n",
       "      <th>respiratory_rate</th>\n",
       "      <th>temperature</th>\n",
       "      <th>wbc</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>60.250000</td>\n",
       "      <td>77.250000</td>\n",
       "      <td>117.333333</td>\n",
       "      <td>77.125000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>92.250000</td>\n",
       "      <td>15.583333</td>\n",
       "      <td>36.166667</td>\n",
       "      <td>5.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-05</td>\n",
       "      <td>48.0</td>\n",
       "      <td>51.782609</td>\n",
       "      <td>64.000000</td>\n",
       "      <td>102.652174</td>\n",
       "      <td>78.906250</td>\n",
       "      <td>95.000000</td>\n",
       "      <td>92.375000</td>\n",
       "      <td>16.782609</td>\n",
       "      <td>36.472222</td>\n",
       "      <td>6.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-06</td>\n",
       "      <td>NaN</td>\n",
       "      <td>50.304348</td>\n",
       "      <td>66.125000</td>\n",
       "      <td>108.391304</td>\n",
       "      <td>82.115385</td>\n",
       "      <td>75.714286</td>\n",
       "      <td>92.538462</td>\n",
       "      <td>17.333333</td>\n",
       "      <td>36.657407</td>\n",
       "      <td>5.3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-07</td>\n",
       "      <td>NaN</td>\n",
       "      <td>57.083333</td>\n",
       "      <td>71.125000</td>\n",
       "      <td>117.208333</td>\n",
       "      <td>82.620690</td>\n",
       "      <td>95.000000</td>\n",
       "      <td>93.960000</td>\n",
       "      <td>15.840000</td>\n",
       "      <td>36.129630</td>\n",
       "      <td>5.4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-08</td>\n",
       "      <td>NaN</td>\n",
       "      <td>53.956522</td>\n",
       "      <td>66.833333</td>\n",
       "      <td>108.291667</td>\n",
       "      <td>82.172414</td>\n",
       "      <td>95.000000</td>\n",
       "      <td>93.500000</td>\n",
       "      <td>15.478261</td>\n",
       "      <td>35.944444</td>\n",
       "      <td>3.8</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17237</th>\n",
       "      <td>19992885</td>\n",
       "      <td>20352341</td>\n",
       "      <td>35180845</td>\n",
       "      <td>2152-07-22</td>\n",
       "      <td>31.0</td>\n",
       "      <td>57.521739</td>\n",
       "      <td>82.173913</td>\n",
       "      <td>136.391304</td>\n",
       "      <td>78.344828</td>\n",
       "      <td>38.333333</td>\n",
       "      <td>99.086957</td>\n",
       "      <td>12.789474</td>\n",
       "      <td>37.055556</td>\n",
       "      <td>12.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17238</th>\n",
       "      <td>19992885</td>\n",
       "      <td>20352341</td>\n",
       "      <td>35180845</td>\n",
       "      <td>2152-07-23</td>\n",
       "      <td>33.0</td>\n",
       "      <td>62.434783</td>\n",
       "      <td>90.217391</td>\n",
       "      <td>149.086957</td>\n",
       "      <td>74.964286</td>\n",
       "      <td>35.000000</td>\n",
       "      <td>98.461538</td>\n",
       "      <td>14.657895</td>\n",
       "      <td>37.296296</td>\n",
       "      <td>15.7</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17239</th>\n",
       "      <td>19992885</td>\n",
       "      <td>20352341</td>\n",
       "      <td>35180845</td>\n",
       "      <td>2152-07-24</td>\n",
       "      <td>NaN</td>\n",
       "      <td>67.838710</td>\n",
       "      <td>97.161290</td>\n",
       "      <td>151.000000</td>\n",
       "      <td>72.303030</td>\n",
       "      <td>35.000000</td>\n",
       "      <td>91.687500</td>\n",
       "      <td>19.195122</td>\n",
       "      <td>36.972222</td>\n",
       "      <td>17.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17240</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-24</td>\n",
       "      <td>37.4</td>\n",
       "      <td>65.937500</td>\n",
       "      <td>80.125000</td>\n",
       "      <td>108.062500</td>\n",
       "      <td>71.000000</td>\n",
       "      <td>48.000000</td>\n",
       "      <td>99.133333</td>\n",
       "      <td>14.600000</td>\n",
       "      <td>36.361111</td>\n",
       "      <td>15.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17241</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-25</td>\n",
       "      <td>NaN</td>\n",
       "      <td>55.571429</td>\n",
       "      <td>69.285714</td>\n",
       "      <td>102.571429</td>\n",
       "      <td>73.583333</td>\n",
       "      <td>50.000000</td>\n",
       "      <td>98.476190</td>\n",
       "      <td>16.954545</td>\n",
       "      <td>36.700000</td>\n",
       "      <td>12.8</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>17242 rows × 14 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "final_label  subject_id   hadm_id   stay_id        date  \\\n",
       "0              10002155  23822395  33685454  2129-08-04   \n",
       "1              10002155  23822395  33685454  2129-08-05   \n",
       "2              10002155  23822395  33685454  2129-08-06   \n",
       "3              10002155  23822395  33685454  2129-08-07   \n",
       "4              10002155  23822395  33685454  2129-08-08   \n",
       "...                 ...       ...       ...         ...   \n",
       "17237          19992885  20352341  35180845  2152-07-22   \n",
       "17238          19992885  20352341  35180845  2152-07-23   \n",
       "17239          19992885  20352341  35180845  2152-07-24   \n",
       "17240          19995091  27602139  32457489  2172-08-24   \n",
       "17241          19995091  27602139  32457489  2172-08-25   \n",
       "\n",
       "final_label  arterial_co2_pressure  blood_pressure_diastolic  \\\n",
       "0                              NaN                 60.250000   \n",
       "1                             48.0                 51.782609   \n",
       "2                              NaN                 50.304348   \n",
       "3                              NaN                 57.083333   \n",
       "4                              NaN                 53.956522   \n",
       "...                            ...                       ...   \n",
       "17237                         31.0                 57.521739   \n",
       "17238                         33.0                 62.434783   \n",
       "17239                          NaN                 67.838710   \n",
       "17240                         37.4                 65.937500   \n",
       "17241                          NaN                 55.571429   \n",
       "\n",
       "final_label  blood_pressure_mean  blood_pressure_systolic  heart_rate  \\\n",
       "0                      77.250000               117.333333   77.125000   \n",
       "1                      64.000000               102.652174   78.906250   \n",
       "2                      66.125000               108.391304   82.115385   \n",
       "3                      71.125000               117.208333   82.620690   \n",
       "4                      66.833333               108.291667   82.172414   \n",
       "...                          ...                      ...         ...   \n",
       "17237                  82.173913               136.391304   78.344828   \n",
       "17238                  90.217391               149.086957   74.964286   \n",
       "17239                  97.161290               151.000000   72.303030   \n",
       "17240                  80.125000               108.062500   71.000000   \n",
       "17241                  69.285714               102.571429   73.583333   \n",
       "\n",
       "final_label  inspired_o2_fraction  o2_saturation_pulseoxymetry  \\\n",
       "0                             NaN                    92.250000   \n",
       "1                       95.000000                    92.375000   \n",
       "2                       75.714286                    92.538462   \n",
       "3                       95.000000                    93.960000   \n",
       "4                       95.000000                    93.500000   \n",
       "...                           ...                          ...   \n",
       "17237                   38.333333                    99.086957   \n",
       "17238                   35.000000                    98.461538   \n",
       "17239                   35.000000                    91.687500   \n",
       "17240                   48.000000                    99.133333   \n",
       "17241                   50.000000                    98.476190   \n",
       "\n",
       "final_label  respiratory_rate  temperature   wbc  \n",
       "0                   15.583333    36.166667   5.5  \n",
       "1                   16.782609    36.472222   6.5  \n",
       "2                   17.333333    36.657407   5.3  \n",
       "3                   15.840000    36.129630   5.4  \n",
       "4                   15.478261    35.944444   3.8  \n",
       "...                       ...          ...   ...  \n",
       "17237               12.789474    37.055556  12.5  \n",
       "17238               14.657895    37.296296  15.7  \n",
       "17239               19.195122    36.972222  17.1  \n",
       "17240               14.600000    36.361111  15.0  \n",
       "17241               16.954545    36.700000  12.8  \n",
       "\n",
       "[17242 rows x 14 columns]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Pivot\n",
    "catch22_pivoted2 = pd.pivot_table(selected_catch22, index=['subject_id', 'hadm_id', 'stay_id', 'date'], columns=['final_label'], values=['valuenum']) # Removed hour from pivot as dont need as not doiing c22\n",
    "catch22_pivoted2.columns = catch22_pivoted2.columns.droplevel()\n",
    "catch22_pivoted2 = catch22_pivoted2.reset_index()\n",
    "catch22_pivoted2.columns = catch22_pivoted2.columns.str.replace(' ','_') \n",
    "catch22_pivoted2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average % of missing values (columns): 0.37030526007427544\n",
      "Average % of missing values (row): 0.37030526007363374\n"
     ]
    }
   ],
   "source": [
    "print('Average % of missing values (columns):', (catch22_pivoted2.isnull().sum(axis=0)/len(catch22_pivoted2)).mean())\n",
    "print('Average % of missing values (row):', (catch22_pivoted2.isnull().sum(axis=1)/len(catch22_pivoted2.columns)).mean())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Rest of features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert to datetime \n",
    "filtered_chartevents3['charttime'] =  pd.to_datetime(filtered_chartevents3['charttime']).dt.date\n",
    "# Rename\n",
    "filtered_chartevents3.rename(columns={'charttime':'date'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.000000    1963\n",
       "1.000000     930\n",
       "0.500000     224\n",
       "0.333333     198\n",
       "0.166667     165\n",
       "0.666667     155\n",
       "0.833333     101\n",
       "0.250000     101\n",
       "0.200000     100\n",
       "0.400000      62\n",
       "0.750000      54\n",
       "0.600000      53\n",
       "0.800000      42\n",
       "0.142857      21\n",
       "0.285714      21\n",
       "0.571429      21\n",
       "0.428571      18\n",
       "0.857143      14\n",
       "0.714286      11\n",
       "0.625000       9\n",
       "0.875000       9\n",
       "0.125000       5\n",
       "0.222222       5\n",
       "0.777778       4\n",
       "0.375000       4\n",
       "0.545455       3\n",
       "0.444444       3\n",
       "0.454545       2\n",
       "0.888889       2\n",
       "0.692308       2\n",
       "0.916667       2\n",
       "0.900000       2\n",
       "0.363636       2\n",
       "0.272727       2\n",
       "0.416667       2\n",
       "0.818182       2\n",
       "0.923077       2\n",
       "0.100000       2\n",
       "0.230769       1\n",
       "0.700000       1\n",
       "0.809524       1\n",
       "0.153846       1\n",
       "0.909091       1\n",
       "0.526316       1\n",
       "0.130435       1\n",
       "0.882353       1\n",
       "0.352941       1\n",
       "0.933333       1\n",
       "0.050000       1\n",
       "0.083333       1\n",
       "0.647059       1\n",
       "0.181818       1\n",
       "0.538462       1\n",
       "0.300000       1\n",
       "0.954545       1\n",
       "0.727273       1\n",
       "0.928571       1\n",
       "0.944444       1\n",
       "0.357143       1\n",
       "Name: level_of_consciousness, dtype: int64"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# level of consciousness\n",
    "consciousness = filtered_chartevents3[filtered_chartevents3['final_label'] == 'level of consciousness']\n",
    "#consciousness\n",
    "consciousness.value = consciousness.value.map({'Alert':0, 'Arouse to Voice':1, 'Lethargic':1, 'Arouse to Pain':1, 'Unresponsive':1, 'Arouse to Stimulation':1, 'Awake/Unresponsive':1})\n",
    "#consciousness = consciousness.drop_duplicates(subset=(['subject_id', 'hadm_id', 'date', 'final_label', 'value']))\n",
    "#consciousness\n",
    "\n",
    "# Pivot\n",
    "consciousness_pivoted = pd.pivot_table(consciousness, index=['subject_id', 'hadm_id', 'stay_id', 'date'], columns=['final_label'], values=['value'])\n",
    "# Drop First Level and add level label as prefix\n",
    "consciousness_pivoted.columns = consciousness_pivoted.columns.droplevel()\n",
    "consciousness_pivoted = consciousness_pivoted.reset_index()\n",
    "consciousness_pivoted.columns = consciousness_pivoted.columns.str.replace(' ','_') \n",
    "\n",
    "# Note left so that you can have a range from 0 to 1 for this (mean of day)\n",
    "consciousness_pivoted.level_of_consciousness.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# GCS and Richmond\n",
    "consciousness = filtered_chartevents3[(filtered_chartevents3['final_label'].str.contains('gcs'))|(filtered_chartevents3['final_label'].str.contains('richmond'))]\n",
    "# Pivot\n",
    "consciousness_pivoted2 = pd.pivot_table(consciousness, index=['subject_id', 'hadm_id', 'stay_id', 'date'], columns=['final_label'], values=['valuenum'])\n",
    "# Drop First Level and add level label as prefix\n",
    "consciousness_pivoted2.columns = consciousness_pivoted2.columns.droplevel()\n",
    "consciousness_pivoted2 = consciousness_pivoted2.reset_index()\n",
    "consciousness_pivoted2.columns = consciousness_pivoted2.columns.str.replace(' ','_') \n",
    "consciousness_pivoted2.rename(columns={'gcs_-_eye_opening': 'gcs_eye_opening', 'gcs_-_motor_response':'gcs_motor_response',\t'gcs_-_verbal_response':'gcs_verbal_response',\t'richmond-ras_scale':'richmond_ras_scale'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "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>final_label</th>\n",
       "      <th>subject_id</th>\n",
       "      <th>hadm_id</th>\n",
       "      <th>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>gcs_eye_opening</th>\n",
       "      <th>gcs_motor_response</th>\n",
       "      <th>gcs_verbal_response</th>\n",
       "      <th>richmond_ras_scale</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-04</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-05</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-06</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-07</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>10002155</td>\n",
       "      <td>23822395</td>\n",
       "      <td>33685454</td>\n",
       "      <td>2129-08-08</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>NaN</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16842</th>\n",
       "      <td>19992885</td>\n",
       "      <td>20352341</td>\n",
       "      <td>35180845</td>\n",
       "      <td>2152-07-22</td>\n",
       "      <td>2.133333</td>\n",
       "      <td>5.533333</td>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16843</th>\n",
       "      <td>19992885</td>\n",
       "      <td>20352341</td>\n",
       "      <td>35180845</td>\n",
       "      <td>2152-07-23</td>\n",
       "      <td>1.833333</td>\n",
       "      <td>4.333333</td>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16844</th>\n",
       "      <td>19992885</td>\n",
       "      <td>20352341</td>\n",
       "      <td>35180845</td>\n",
       "      <td>2152-07-24</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>1.0</td>\n",
       "      <td>-1.000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16845</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-24</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>3.0</td>\n",
       "      <td>-0.250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16846</th>\n",
       "      <td>19995091</td>\n",
       "      <td>27602139</td>\n",
       "      <td>32457489</td>\n",
       "      <td>2172-08-25</td>\n",
       "      <td>3.250000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.0</td>\n",
       "      <td>0.000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>16847 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "final_label  subject_id   hadm_id   stay_id        date  gcs_eye_opening  \\\n",
       "0              10002155  23822395  33685454  2129-08-04         4.000000   \n",
       "1              10002155  23822395  33685454  2129-08-05         4.000000   \n",
       "2              10002155  23822395  33685454  2129-08-06         4.000000   \n",
       "3              10002155  23822395  33685454  2129-08-07         4.000000   \n",
       "4              10002155  23822395  33685454  2129-08-08         4.000000   \n",
       "...                 ...       ...       ...         ...              ...   \n",
       "16842          19992885  20352341  35180845  2152-07-22         2.133333   \n",
       "16843          19992885  20352341  35180845  2152-07-23         1.833333   \n",
       "16844          19992885  20352341  35180845  2152-07-24         2.000000   \n",
       "16845          19995091  27602139  32457489  2172-08-24         2.000000   \n",
       "16846          19995091  27602139  32457489  2172-08-25         3.250000   \n",
       "\n",
       "final_label  gcs_motor_response  gcs_verbal_response  richmond_ras_scale  \n",
       "0                      6.000000                  5.0                 NaN  \n",
       "1                      6.000000                  5.0                 NaN  \n",
       "2                      6.000000                  5.0                 NaN  \n",
       "3                      6.000000                  5.0                 NaN  \n",
       "4                      6.000000                  5.0                 NaN  \n",
       "...                         ...                  ...                 ...  \n",
       "16842                  5.533333                  1.0              -1.250  \n",
       "16843                  4.333333                  1.0              -1.375  \n",
       "16844                  4.000000                  1.0              -1.000  \n",
       "16845                  3.500000                  3.0              -0.250  \n",
       "16846                  6.000000                  5.0               0.000  \n",
       "\n",
       "[16847 rows x 8 columns]"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "consciousness_pivoted2"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "NEWS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Merge pivots\n",
    "chartevents_pivoted = pd.merge(catch22_pivoted2, consciousness_pivoted, how='outer') # on=['stay_id', 'date']\n",
    "chartevents_pivoted = pd.merge(chartevents_pivoted, consciousness_pivoted2, how='outer')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 89,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4347"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "4347"
      ]
     },
     "execution_count": 89,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hold_out_data.stay_id.nunique()\n",
    "len(set(chartevents_pivoted.stay_id.unique()) & set(hold_out_data.stay_id.unique()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert to str or does not work\n",
    "hold_out_data.date = hold_out_data.date.astype(str)\n",
    "chartevents_pivoted.date = chartevents_pivoted.date.astype(str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "chartevents_pivoted = pd.merge(hold_out_data, chartevents_pivoted.drop(columns=['subject_id', 'hadm_id']), how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "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>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>iv_flag</th>\n",
       "      <th>first_po_flag</th>\n",
       "      <th>po_flag</th>\n",
       "      <th>iv_treatment_length</th>\n",
       "      <th>arterial_co2_pressure</th>\n",
       "      <th>blood_pressure_diastolic</th>\n",
       "      <th>blood_pressure_mean</th>\n",
       "      <th>blood_pressure_systolic</th>\n",
       "      <th>...</th>\n",
       "      <th>inspired_o2_fraction</th>\n",
       "      <th>o2_saturation_pulseoxymetry</th>\n",
       "      <th>respiratory_rate</th>\n",
       "      <th>temperature</th>\n",
       "      <th>wbc</th>\n",
       "      <th>level_of_consciousness</th>\n",
       "      <th>gcs_eye_opening</th>\n",
       "      <th>gcs_motor_response</th>\n",
       "      <th>gcs_verbal_response</th>\n",
       "      <th>richmond_ras_scale</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-11</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>56.300000</td>\n",
       "      <td>78.900000</td>\n",
       "      <td>117.300000</td>\n",
       "      <td>...</td>\n",
       "      <td>63.333333</td>\n",
       "      <td>99.100000</td>\n",
       "      <td>11.080000</td>\n",
       "      <td>36.488889</td>\n",
       "      <td>7.90</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.666667</td>\n",
       "      <td>4.333333</td>\n",
       "      <td>2.333333</td>\n",
       "      <td>-1.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-12</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>54.840000</td>\n",
       "      <td>75.600000</td>\n",
       "      <td>121.040000</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.541667</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>37.400855</td>\n",
       "      <td>13.30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>-0.090909</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-13</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>72.909091</td>\n",
       "      <td>81.363636</td>\n",
       "      <td>107.272727</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.363636</td>\n",
       "      <td>20.272727</td>\n",
       "      <td>36.916667</td>\n",
       "      <td>15.10</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-21</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>75.000000</td>\n",
       "      <td>87.806452</td>\n",
       "      <td>115.225806</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.000000</td>\n",
       "      <td>26.580645</td>\n",
       "      <td>37.694444</td>\n",
       "      <td>11.00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-22</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>69.142857</td>\n",
       "      <td>81.095238</td>\n",
       "      <td>106.619048</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>98.333333</td>\n",
       "      <td>22.285714</td>\n",
       "      <td>37.333333</td>\n",
       "      <td>11.45</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</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>14298</th>\n",
       "      <td>37914144</td>\n",
       "      <td>2136-05-23</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.235294</td>\n",
       "      <td>93.647059</td>\n",
       "      <td>125.058824</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>94.944444</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>36.677778</td>\n",
       "      <td>13.30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14299</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>28.250000</td>\n",
       "      <td>39.750000</td>\n",
       "      <td>77.750000</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>94.600000</td>\n",
       "      <td>25.400000</td>\n",
       "      <td>37.611111</td>\n",
       "      <td>15.60</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14300</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>34.500000</td>\n",
       "      <td>46.700000</td>\n",
       "      <td>88.600000</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>93.095238</td>\n",
       "      <td>24.863636</td>\n",
       "      <td>37.031746</td>\n",
       "      <td>13.85</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14301</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-19</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>45.304348</td>\n",
       "      <td>56.391304</td>\n",
       "      <td>93.739130</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.956522</td>\n",
       "      <td>22.958333</td>\n",
       "      <td>36.788889</td>\n",
       "      <td>10.80</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14302</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-20</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>54.578947</td>\n",
       "      <td>67.736842</td>\n",
       "      <td>112.736842</td>\n",
       "      <td>...</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.000000</td>\n",
       "      <td>25.210526</td>\n",
       "      <td>36.930556</td>\n",
       "      <td>10.80</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>14303 rows × 21 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        stay_id        date  iv_flag  first_po_flag  po_flag  \\\n",
       "0      31658148  2154-01-11      1.0            1.0        0   \n",
       "1      31658148  2154-01-12      1.0            1.0        0   \n",
       "2      31658148  2154-01-13      1.0            1.0        0   \n",
       "3      33816676  2117-08-21      1.0            1.0        0   \n",
       "4      33816676  2117-08-22      1.0            1.0        0   \n",
       "...         ...         ...      ...            ...      ...   \n",
       "14298  37914144  2136-05-23      1.0            1.0        0   \n",
       "14299  37246494  2127-07-17      1.0            1.0        0   \n",
       "14300  37246494  2127-07-18      1.0            1.0        0   \n",
       "14301  37246494  2127-07-19      1.0            1.0        0   \n",
       "14302  37246494  2127-07-20      1.0            1.0        0   \n",
       "\n",
       "       iv_treatment_length  arterial_co2_pressure  blood_pressure_diastolic  \\\n",
       "0                        0                   37.0                 56.300000   \n",
       "1                        1                    NaN                 54.840000   \n",
       "2                        2                    NaN                 72.909091   \n",
       "3                        0                   39.0                 75.000000   \n",
       "4                        1                    NaN                 69.142857   \n",
       "...                    ...                    ...                       ...   \n",
       "14298                    1                    NaN                 76.235294   \n",
       "14299                    0                    NaN                 28.250000   \n",
       "14300                    1                    NaN                 34.500000   \n",
       "14301                    2                    NaN                 45.304348   \n",
       "14302                    3                    NaN                 54.578947   \n",
       "\n",
       "       blood_pressure_mean  blood_pressure_systolic  ...  \\\n",
       "0                78.900000               117.300000  ...   \n",
       "1                75.600000               121.040000  ...   \n",
       "2                81.363636               107.272727  ...   \n",
       "3                87.806452               115.225806  ...   \n",
       "4                81.095238               106.619048  ...   \n",
       "...                    ...                      ...  ...   \n",
       "14298            93.647059               125.058824  ...   \n",
       "14299            39.750000                77.750000  ...   \n",
       "14300            46.700000                88.600000  ...   \n",
       "14301            56.391304                93.739130  ...   \n",
       "14302            67.736842               112.736842  ...   \n",
       "\n",
       "       inspired_o2_fraction  o2_saturation_pulseoxymetry  respiratory_rate  \\\n",
       "0                 63.333333                    99.100000         11.080000   \n",
       "1                       NaN                    96.541667         15.000000   \n",
       "2                       NaN                    95.363636         20.272727   \n",
       "3                       NaN                    97.000000         26.580645   \n",
       "4                       NaN                    98.333333         22.285714   \n",
       "...                     ...                          ...               ...   \n",
       "14298                   NaN                    94.944444         16.000000   \n",
       "14299                   NaN                    94.600000         25.400000   \n",
       "14300                   NaN                    93.095238         24.863636   \n",
       "14301                   NaN                    95.956522         22.958333   \n",
       "14302                   NaN                    96.000000         25.210526   \n",
       "\n",
       "       temperature    wbc  level_of_consciousness  gcs_eye_opening  \\\n",
       "0        36.488889   7.90                     NaN         2.666667   \n",
       "1        37.400855  13.30                     NaN         4.000000   \n",
       "2        36.916667  15.10                     NaN         4.000000   \n",
       "3        37.694444  11.00                     NaN         4.000000   \n",
       "4        37.333333  11.45                     NaN         4.000000   \n",
       "...            ...    ...                     ...              ...   \n",
       "14298    36.677778  13.30                     NaN         4.000000   \n",
       "14299    37.611111  15.60                     NaN         4.000000   \n",
       "14300    37.031746  13.85                     NaN         4.000000   \n",
       "14301    36.788889  10.80                     NaN         4.000000   \n",
       "14302    36.930556  10.80                     NaN         4.000000   \n",
       "\n",
       "       gcs_motor_response  gcs_verbal_response  richmond_ras_scale  \n",
       "0                4.333333             2.333333           -1.000000  \n",
       "1                6.000000             5.000000           -0.090909  \n",
       "2                6.000000             5.000000            0.000000  \n",
       "3                6.000000             5.000000            0.000000  \n",
       "4                6.000000             5.000000            0.000000  \n",
       "...                   ...                  ...                 ...  \n",
       "14298            6.000000             5.000000            0.000000  \n",
       "14299            6.000000             5.000000            0.000000  \n",
       "14300            6.000000             5.000000            0.000000  \n",
       "14301            6.000000             5.000000            0.000000  \n",
       "14302            6.000000             5.000000            0.000000  \n",
       "\n",
       "[14303 rows x 21 columns]"
      ]
     },
     "execution_count": 119,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chartevents_pivoted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "def NEWS2_function(respiratory_rate, arterial_co2_pressure, o2_saturation_pulseoxymetry, inspired_o2_fraction, temperature, blood_pressure_systolic, heart_rate, level_of_consciousness, gcs_eye_opening, gcs_motor_response, gcs_verbal_response, richmond_ras_scale):\n",
    "    score = 0\n",
    "\n",
    "    if respiratory_rate <= 8:\n",
    "        score += 3\n",
    "    elif respiratory_rate < 12:\n",
    "        score += 1\n",
    "    elif respiratory_rate >= 25:\n",
    "        score += 3\n",
    "    elif respiratory_rate > 20:\n",
    "        score += 2\n",
    "    \n",
    "    if arterial_co2_pressure >= 50:\n",
    "        if o2_saturation_pulseoxymetry <= 83:\n",
    "            score += 3\n",
    "        elif o2_saturation_pulseoxymetry < 86:\n",
    "            score += 2\n",
    "        elif o2_saturation_pulseoxymetry < 88:\n",
    "            score += 1\n",
    "        elif inspired_o2_fraction > 21:\n",
    "            if o2_saturation_pulseoxymetry >= 97:\n",
    "                score += 3\n",
    "            elif o2_saturation_pulseoxymetry > 95:\n",
    "                score += 2\n",
    "            elif o2_saturation_pulseoxymetry > 93:\n",
    "                score += 1\n",
    "    else:\n",
    "        if o2_saturation_pulseoxymetry <= 91:\n",
    "            score += 3\n",
    "        elif o2_saturation_pulseoxymetry < 94:\n",
    "            score += 2\n",
    "        elif o2_saturation_pulseoxymetry < 96:\n",
    "            score += 1\n",
    "\n",
    "    if inspired_o2_fraction > 21:\n",
    "        score += 2\n",
    "    \n",
    "    if temperature <= 35:\n",
    "        score += 3\n",
    "    elif temperature <= 36:\n",
    "        score += 1\n",
    "    elif temperature > 39:\n",
    "        score += 2\n",
    "    elif temperature > 38:\n",
    "        score += 1\n",
    "    \n",
    "    if blood_pressure_systolic <= 90:\n",
    "        score += 3\n",
    "    elif blood_pressure_systolic <= 100:\n",
    "        score += 2\n",
    "    elif blood_pressure_systolic <= 100:\n",
    "        score += 1\n",
    "    elif blood_pressure_systolic >= 220:\n",
    "        score += 3\n",
    "    \n",
    "    if heart_rate <= 40:\n",
    "        score += 3\n",
    "    elif heart_rate <= 50:\n",
    "        score += 1\n",
    "    elif heart_rate >= 131:\n",
    "        score += 3\n",
    "    elif heart_rate >= 111:\n",
    "        score += 2\n",
    "    elif heart_rate >= 91:\n",
    "        score += 1\n",
    "    \n",
    "    if level_of_consciousness > 0:\n",
    "        score += 3\n",
    "    elif gcs_eye_opening < 4:\n",
    "        score += 3\n",
    "    elif gcs_verbal_response < 5:\n",
    "        score += 3\n",
    "    elif gcs_motor_response < 6:\n",
    "        score += 3\n",
    "    elif richmond_ras_scale > 0:\n",
    "        score += 3\n",
    "    elif richmond_ras_scale < 0:\n",
    "        score += 3\n",
    "\n",
    "    return score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate NEWS 2 score for each\n",
    "chartevents_pivoted['news2'] = chartevents_pivoted.apply(lambda x: NEWS2_function(x['respiratory_rate'], x['arterial_co2_pressure'], x['o2_saturation_pulseoxymetry'], x['inspired_o2_fraction'], x['temperature'], x['blood_pressure_systolic'], x['heart_rate'], x['level_of_consciousness'], x['gcs_eye_opening'], x['gcs_motor_response'], x['gcs_verbal_response'], x['richmond_ras_scale']), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "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>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>iv_flag</th>\n",
       "      <th>first_po_flag</th>\n",
       "      <th>po_flag</th>\n",
       "      <th>iv_treatment_length</th>\n",
       "      <th>arterial_co2_pressure</th>\n",
       "      <th>blood_pressure_diastolic</th>\n",
       "      <th>blood_pressure_mean</th>\n",
       "      <th>blood_pressure_systolic</th>\n",
       "      <th>...</th>\n",
       "      <th>o2_saturation_pulseoxymetry</th>\n",
       "      <th>respiratory_rate</th>\n",
       "      <th>temperature</th>\n",
       "      <th>wbc</th>\n",
       "      <th>level_of_consciousness</th>\n",
       "      <th>gcs_eye_opening</th>\n",
       "      <th>gcs_motor_response</th>\n",
       "      <th>gcs_verbal_response</th>\n",
       "      <th>richmond_ras_scale</th>\n",
       "      <th>news2</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-11</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>56.300000</td>\n",
       "      <td>78.900000</td>\n",
       "      <td>117.300000</td>\n",
       "      <td>...</td>\n",
       "      <td>99.100000</td>\n",
       "      <td>11.080000</td>\n",
       "      <td>36.488889</td>\n",
       "      <td>7.90</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.666667</td>\n",
       "      <td>4.333333</td>\n",
       "      <td>2.333333</td>\n",
       "      <td>-1.000000</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-12</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>54.840000</td>\n",
       "      <td>75.600000</td>\n",
       "      <td>121.040000</td>\n",
       "      <td>...</td>\n",
       "      <td>96.541667</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>37.400855</td>\n",
       "      <td>13.30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>-0.090909</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-13</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>72.909091</td>\n",
       "      <td>81.363636</td>\n",
       "      <td>107.272727</td>\n",
       "      <td>...</td>\n",
       "      <td>95.363636</td>\n",
       "      <td>20.272727</td>\n",
       "      <td>36.916667</td>\n",
       "      <td>15.10</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-21</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>39.0</td>\n",
       "      <td>75.000000</td>\n",
       "      <td>87.806452</td>\n",
       "      <td>115.225806</td>\n",
       "      <td>...</td>\n",
       "      <td>97.000000</td>\n",
       "      <td>26.580645</td>\n",
       "      <td>37.694444</td>\n",
       "      <td>11.00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-22</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>69.142857</td>\n",
       "      <td>81.095238</td>\n",
       "      <td>106.619048</td>\n",
       "      <td>...</td>\n",
       "      <td>98.333333</td>\n",
       "      <td>22.285714</td>\n",
       "      <td>37.333333</td>\n",
       "      <td>11.45</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3</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>14298</th>\n",
       "      <td>37914144</td>\n",
       "      <td>2136-05-23</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.235294</td>\n",
       "      <td>93.647059</td>\n",
       "      <td>125.058824</td>\n",
       "      <td>...</td>\n",
       "      <td>94.944444</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>36.677778</td>\n",
       "      <td>13.30</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14299</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-17</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>28.250000</td>\n",
       "      <td>39.750000</td>\n",
       "      <td>77.750000</td>\n",
       "      <td>...</td>\n",
       "      <td>94.600000</td>\n",
       "      <td>25.400000</td>\n",
       "      <td>37.611111</td>\n",
       "      <td>15.60</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14300</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>34.500000</td>\n",
       "      <td>46.700000</td>\n",
       "      <td>88.600000</td>\n",
       "      <td>...</td>\n",
       "      <td>93.095238</td>\n",
       "      <td>24.863636</td>\n",
       "      <td>37.031746</td>\n",
       "      <td>13.85</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14301</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-19</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>45.304348</td>\n",
       "      <td>56.391304</td>\n",
       "      <td>93.739130</td>\n",
       "      <td>...</td>\n",
       "      <td>95.956522</td>\n",
       "      <td>22.958333</td>\n",
       "      <td>36.788889</td>\n",
       "      <td>10.80</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>6</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14302</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-20</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>54.578947</td>\n",
       "      <td>67.736842</td>\n",
       "      <td>112.736842</td>\n",
       "      <td>...</td>\n",
       "      <td>96.000000</td>\n",
       "      <td>25.210526</td>\n",
       "      <td>36.930556</td>\n",
       "      <td>10.80</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>14303 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        stay_id        date  iv_flag  first_po_flag  po_flag  \\\n",
       "0      31658148  2154-01-11      1.0            1.0        0   \n",
       "1      31658148  2154-01-12      1.0            1.0        0   \n",
       "2      31658148  2154-01-13      1.0            1.0        0   \n",
       "3      33816676  2117-08-21      1.0            1.0        0   \n",
       "4      33816676  2117-08-22      1.0            1.0        0   \n",
       "...         ...         ...      ...            ...      ...   \n",
       "14298  37914144  2136-05-23      1.0            1.0        0   \n",
       "14299  37246494  2127-07-17      1.0            1.0        0   \n",
       "14300  37246494  2127-07-18      1.0            1.0        0   \n",
       "14301  37246494  2127-07-19      1.0            1.0        0   \n",
       "14302  37246494  2127-07-20      1.0            1.0        0   \n",
       "\n",
       "       iv_treatment_length  arterial_co2_pressure  blood_pressure_diastolic  \\\n",
       "0                        0                   37.0                 56.300000   \n",
       "1                        1                    NaN                 54.840000   \n",
       "2                        2                    NaN                 72.909091   \n",
       "3                        0                   39.0                 75.000000   \n",
       "4                        1                    NaN                 69.142857   \n",
       "...                    ...                    ...                       ...   \n",
       "14298                    1                    NaN                 76.235294   \n",
       "14299                    0                    NaN                 28.250000   \n",
       "14300                    1                    NaN                 34.500000   \n",
       "14301                    2                    NaN                 45.304348   \n",
       "14302                    3                    NaN                 54.578947   \n",
       "\n",
       "       blood_pressure_mean  blood_pressure_systolic  ...  \\\n",
       "0                78.900000               117.300000  ...   \n",
       "1                75.600000               121.040000  ...   \n",
       "2                81.363636               107.272727  ...   \n",
       "3                87.806452               115.225806  ...   \n",
       "4                81.095238               106.619048  ...   \n",
       "...                    ...                      ...  ...   \n",
       "14298            93.647059               125.058824  ...   \n",
       "14299            39.750000                77.750000  ...   \n",
       "14300            46.700000                88.600000  ...   \n",
       "14301            56.391304                93.739130  ...   \n",
       "14302            67.736842               112.736842  ...   \n",
       "\n",
       "       o2_saturation_pulseoxymetry  respiratory_rate  temperature    wbc  \\\n",
       "0                        99.100000         11.080000    36.488889   7.90   \n",
       "1                        96.541667         15.000000    37.400855  13.30   \n",
       "2                        95.363636         20.272727    36.916667  15.10   \n",
       "3                        97.000000         26.580645    37.694444  11.00   \n",
       "4                        98.333333         22.285714    37.333333  11.45   \n",
       "...                            ...               ...          ...    ...   \n",
       "14298                    94.944444         16.000000    36.677778  13.30   \n",
       "14299                    94.600000         25.400000    37.611111  15.60   \n",
       "14300                    93.095238         24.863636    37.031746  13.85   \n",
       "14301                    95.956522         22.958333    36.788889  10.80   \n",
       "14302                    96.000000         25.210526    36.930556  10.80   \n",
       "\n",
       "       level_of_consciousness  gcs_eye_opening  gcs_motor_response  \\\n",
       "0                         NaN         2.666667            4.333333   \n",
       "1                         NaN         4.000000            6.000000   \n",
       "2                         NaN         4.000000            6.000000   \n",
       "3                         NaN         4.000000            6.000000   \n",
       "4                         NaN         4.000000            6.000000   \n",
       "...                       ...              ...                 ...   \n",
       "14298                     NaN         4.000000            6.000000   \n",
       "14299                     NaN         4.000000            6.000000   \n",
       "14300                     NaN         4.000000            6.000000   \n",
       "14301                     NaN         4.000000            6.000000   \n",
       "14302                     NaN         4.000000            6.000000   \n",
       "\n",
       "       gcs_verbal_response  richmond_ras_scale  news2  \n",
       "0                 2.333333           -1.000000      6  \n",
       "1                 5.000000           -0.090909      3  \n",
       "2                 5.000000            0.000000      4  \n",
       "3                 5.000000            0.000000      4  \n",
       "4                 5.000000            0.000000      3  \n",
       "...                    ...                 ...    ...  \n",
       "14298             5.000000            0.000000      1  \n",
       "14299             5.000000            0.000000      8  \n",
       "14300             5.000000            0.000000      8  \n",
       "14301             5.000000            0.000000      6  \n",
       "14302             5.000000            0.000000      3  \n",
       "\n",
       "[14303 rows x 22 columns]"
      ]
     },
     "execution_count": 123,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chartevents_pivoted"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Merge\n",
    "chartevents_pivoted2_2 = pd.merge(chartevents_pivoted, hold_out_data5, how='outer')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "14303\n"
     ]
    }
   ],
   "source": [
    "# Get baseline predictions \n",
    "cumcount = []\n",
    "count = 1\n",
    "pos = -1\n",
    "flag = 0\n",
    "\n",
    "for x in range(len(chartevents_pivoted2_2)):\n",
    "    pos += 1\n",
    "    if pos == 0:\n",
    "        cumcount.append(np.nan)\n",
    "    elif chartevents_pivoted2_2.iloc[x]['stay_id'] == chartevents_pivoted2_2.iloc[x-1]['stay_id']:\n",
    "        if flag == 1:\n",
    "            cumcount.append(1) # Changed to aways predict switch after initial switch prediction for each stay\n",
    "        else:\n",
    "            if np.isnan(chartevents_pivoted2_2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "                cumcount.append(np.nan)\n",
    "            elif chartevents_pivoted2_2.iloc[x]['news2'] < chartevents_pivoted2_2.iloc[x-1]['news2']:\n",
    "                if chartevents_pivoted2_2.iloc[x]['max'] > 38:\n",
    "                    cumcount.append(0)\n",
    "                elif chartevents_pivoted2_2.iloc[x]['min'] < 36:\n",
    "                    cumcount.append(0)\n",
    "                else:\n",
    "                    cumcount.append(1)\n",
    "                    flag = 1\n",
    "            else:\n",
    "                cumcount.append(0)\n",
    "    elif np.isnan(chartevents_pivoted2_2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "        cumcount.append(np.nan)\n",
    "        flag = 0\n",
    "    elif chartevents_pivoted2_2.iloc[x]['news2'] < chartevents_pivoted2_2.iloc[x-1]['news2']:\n",
    "        if chartevents_pivoted2_2.iloc[x]['max'] > 38:\n",
    "            cumcount.append(0)\n",
    "            flag = 0\n",
    "        elif chartevents_pivoted2_2.iloc[x]['min'] < 36:\n",
    "            cumcount.append(0)\n",
    "            flag = 0\n",
    "        else:\n",
    "            cumcount.append(1)\n",
    "            flag = 1\n",
    "    else:\n",
    "        cumcount.append(0)\n",
    "        flag = 0\n",
    "\n",
    "print(len(cumcount))\n",
    "chartevents_pivoted2_2['predictions'] = cumcount"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "14303\n"
     ]
    }
   ],
   "source": [
    "# Get strictbaseline predictions \n",
    "cumcount = []\n",
    "count = 1\n",
    "pos = -1\n",
    "flag = 0\n",
    "\n",
    "for x in range(len(chartevents_pivoted2_2)):\n",
    "    pos += 1\n",
    "    if pos == 0:\n",
    "        cumcount.append(np.nan)\n",
    "    elif chartevents_pivoted2_2.iloc[x]['stay_id'] == chartevents_pivoted2_2.iloc[x-1]['stay_id']:\n",
    "        if flag == 1:\n",
    "            cumcount.append(1) # Changed to aways predict switch after initial switch prediction for each stay\n",
    "        else:\n",
    "            if np.isnan(chartevents_pivoted2_2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "                cumcount.append(np.nan)\n",
    "            elif chartevents_pivoted2_2.iloc[x]['news2'] < chartevents_pivoted2_2.iloc[x-1]['news2']:\n",
    "                if chartevents_pivoted2_2.iloc[x]['news2'] < 7: # Add in threshold of NEWS less than 7\n",
    "                    if chartevents_pivoted2_2.iloc[x]['max'] > 38:\n",
    "                        cumcount.append(0)\n",
    "                    elif chartevents_pivoted2_2.iloc[x]['min'] < 36:\n",
    "                        cumcount.append(0)\n",
    "                    else:\n",
    "                        cumcount.append(1)\n",
    "                        flag = 1\n",
    "                else:\n",
    "                    cumcount.append(0)\n",
    "            else:\n",
    "                cumcount.append(0)\n",
    "    elif np.isnan(chartevents_pivoted2_2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "        cumcount.append(np.nan)\n",
    "        flag = 0\n",
    "    elif chartevents_pivoted2_2.iloc[x]['news2'] < chartevents_pivoted2_2.iloc[x-1]['news2']:\n",
    "        if chartevents_pivoted2_2.iloc[x]['news2'] < 7: # Add in threshold of NEWS less than 7\n",
    "            if chartevents_pivoted2_2.iloc[x]['max'] > 38:\n",
    "                cumcount.append(0)\n",
    "                flag = 0\n",
    "            elif chartevents_pivoted2_2.iloc[x]['min'] < 36:\n",
    "                cumcount.append(0)\n",
    "                flag = 0\n",
    "            else:\n",
    "                cumcount.append(1)\n",
    "                flag = 1\n",
    "        else:\n",
    "            cumcount.append(0)\n",
    "            flag = 0\n",
    "    else:\n",
    "        cumcount.append(0)\n",
    "        flag = 0\n",
    "\n",
    "print(len(cumcount))\n",
    "chartevents_pivoted2_2['strict_predictions'] = cumcount"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 180,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    5480\n",
       "1.0    5459\n",
       "Name: predictions, dtype: int64"
      ]
     },
     "execution_count": 180,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "0.0    6034\n",
       "1.0    4903\n",
       "Name: strict_predictions, dtype: int64"
      ]
     },
     "execution_count": 180,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chartevents_pivoted2_2.predictions.value_counts()\n",
    "chartevents_pivoted2_2.strict_predictions.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get prediction df\n",
    "prediction_df2 = chartevents_pivoted2_2[['stay_id', 'date', 'iv_flag', 'first_po_flag', 'po_flag', 'iv_treatment_length', 'predictions', 'strict_predictions']]\n",
    "prediction_df2 = prediction_df2[prediction_df2['predictions'].notna()]\n",
    "prediction_df2 = prediction_df2[prediction_df2['strict_predictions'].notna()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "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>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>iv_flag</th>\n",
       "      <th>first_po_flag</th>\n",
       "      <th>po_flag</th>\n",
       "      <th>iv_treatment_length</th>\n",
       "      <th>predictions</th>\n",
       "      <th>strict_predictions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-12</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>31658148</td>\n",
       "      <td>2154-01-13</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-21</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>33816676</td>\n",
       "      <td>2117-08-22</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>32889989</td>\n",
       "      <td>2126-06-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14296</th>\n",
       "      <td>30772812</td>\n",
       "      <td>2130-06-08</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14298</th>\n",
       "      <td>37914144</td>\n",
       "      <td>2136-05-23</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14300</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-18</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14301</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-19</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14302</th>\n",
       "      <td>37246494</td>\n",
       "      <td>2127-07-20</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>10937 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        stay_id        date  iv_flag  first_po_flag  po_flag  \\\n",
       "1      31658148  2154-01-12      1.0            1.0        0   \n",
       "2      31658148  2154-01-13      1.0            1.0        0   \n",
       "3      33816676  2117-08-21      1.0            1.0        0   \n",
       "4      33816676  2117-08-22      1.0            1.0        0   \n",
       "6      32889989  2126-06-04      1.0            1.0        0   \n",
       "...         ...         ...      ...            ...      ...   \n",
       "14296  30772812  2130-06-08      1.0            1.0        0   \n",
       "14298  37914144  2136-05-23      1.0            1.0        0   \n",
       "14300  37246494  2127-07-18      1.0            1.0        0   \n",
       "14301  37246494  2127-07-19      1.0            1.0        0   \n",
       "14302  37246494  2127-07-20      1.0            1.0        0   \n",
       "\n",
       "       iv_treatment_length  predictions  strict_predictions  \n",
       "1                        1          0.0                 0.0  \n",
       "2                        2          0.0                 0.0  \n",
       "3                        0          0.0                 0.0  \n",
       "4                        1          0.0                 0.0  \n",
       "6                        1          0.0                 0.0  \n",
       "...                    ...          ...                 ...  \n",
       "14296                    1          0.0                 0.0  \n",
       "14298                    1          1.0                 1.0  \n",
       "14300                    1          0.0                 0.0  \n",
       "14301                    2          1.0                 1.0  \n",
       "14302                    3          1.0                 1.0  \n",
       "\n",
       "[10937 rows x 8 columns]"
      ]
     },
     "execution_count": 186,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "prediction_df2"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUROC: 0.6604952552141343\n",
      "Accuracy: 0.6093992868245406\n",
      "Recall: 0.7505285412262156\n",
      "Precision: 0.3252702950339014\n",
      "F1: 0.45384812068524677\n",
      "AUPRC: 0.2980699632507904\n",
      "CM: [[4890 3682]\n",
      " [ 590 1775]]\n",
      "TPR: 0.7505285412262156\n",
      "FPR: 0.4295380307979468\n"
     ]
    }
   ],
   "source": [
    "get_results(prediction_df2.po_flag, prediction_df2.predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUROC: 0.6596310175516136\n",
      "Accuracy: 0.6375605741976776\n",
      "Recall: 0.6985200845665962\n",
      "Precision: 0.3369365694472772\n",
      "F1: 0.45459548706659325\n",
      "AUPRC: 0.3005485125976789\n",
      "CM: [[5321 3251]\n",
      " [ 713 1652]]\n",
      "TPR: 0.6985200845665962\n",
      "FPR: 0.3792580494633691\n"
     ]
    }
   ],
   "source": [
    "get_results(prediction_df2.po_flag, prediction_df2.strict_predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "treatment_length_df = pd.DataFrame()\n",
    "\n",
    "for i in prediction_df2.iv_treatment_length.unique():\n",
    "    print(i)\n",
    "    temp_test_data = prediction_df2[prediction_df2['iv_treatment_length'] == i]\n",
    "    test_labels = temp_test_data.po_flag\n",
    "    lower_bound_test_predictions = temp_test_data.predictions\n",
    "    upper_bound_test_predictions = temp_test_data.strict_predictions\n",
    "\n",
    "    print(np.unique(test_labels, return_counts=True))\n",
    "    print(np.unique(lower_bound_test_predictions, return_counts=True))\n",
    "\n",
    "    label_values, label_counts = np.unique(test_labels, return_counts=True)\n",
    "    lower_bound_values, lower_bound_counts =np.unique(lower_bound_test_predictions, return_counts=True)\n",
    "    upper_bound_values, upper_bound_counts =np.unique(upper_bound_test_predictions, return_counts=True)\n",
    "\n",
    "    label_0 = 0\n",
    "    label_1 = 0\n",
    "    for x in range(len(label_values)):\n",
    "        if label_values[x] == 0:\n",
    "            label_0 = label_counts[x]\n",
    "        elif label_values[x] == 1:\n",
    "            label_1 = label_counts[x]\n",
    "    prediction_0 = 0\n",
    "    prediction_1 = 0\n",
    "    for x in range(len(lower_bound_values)):\n",
    "        if lower_bound_values[x] == 0:\n",
    "            prediction_0 = lower_bound_counts[x]\n",
    "        elif lower_bound_values[x] == 1:\n",
    "            prediction_1 = lower_bound_counts[x]\n",
    "    ub_prediction_0 = 0\n",
    "    ub_prediction_1 = 0\n",
    "    for x in range(len(upper_bound_values)):\n",
    "        if upper_bound_values[x] == 0:\n",
    "            ub_prediction_0 = upper_bound_counts[x]\n",
    "        elif upper_bound_values[x] == 1:\n",
    "            ub_prediction_1 = upper_bound_counts[x]\n",
    "    \n",
    "    # Lower bound\n",
    "    try:\n",
    "        test_auroc2 = roc_auc_score(test_labels, lower_bound_test_predictions)\n",
    "    except:\n",
    "        test_auroc2 = np.nan\n",
    "    #print('Test AUROC result 2:', test_auroc2)\n",
    "    test_accuracy2 = accuracy_score(test_labels, lower_bound_test_predictions)\n",
    "    #assert test_accuracy == test_accuracy2\n",
    "    test_balanced_accuracy = balanced_accuracy_score(test_labels, lower_bound_test_predictions)\n",
    "    test_recall = recall_score(test_labels, lower_bound_test_predictions)\n",
    "    test_precision = precision_score(test_labels, lower_bound_test_predictions)\n",
    "    test_f1 = f1_score(test_labels, lower_bound_test_predictions)\n",
    "    test_auprc = average_precision_score(test_labels, lower_bound_test_predictions)\n",
    "    test_cm = confusion_matrix(test_labels, lower_bound_test_predictions)\n",
    "    if test_cm.shape == (2, 2):\n",
    "        tn, fp, fn, tp = test_cm.ravel()\n",
    "        test_true_positive_rate = (tp / (tp + fn))\n",
    "        test_false_positive_rate = (fp / (fp + tn))\n",
    "    else:\n",
    "        test_true_positive_rate = np.nan\n",
    "        test_false_positive_rate = np.nan\n",
    "\n",
    "    # Upper bound\n",
    "    try:\n",
    "        ub_test_auroc2 = roc_auc_score(test_labels, upper_bound_test_predictions)\n",
    "    except:\n",
    "        ub_test_auroc2 = np.nan\n",
    "    #print('Test ub_AUROC result 2:', ub_test_auroc2)\n",
    "    ub_test_accuracy2 = accuracy_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_balanced_accuracy = balanced_accuracy_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_recall = recall_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_precision = precision_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_f1 = f1_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_auprc = average_precision_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_cm = confusion_matrix(test_labels, upper_bound_test_predictions)\n",
    "    if ub_test_cm.shape == (2, 2):\n",
    "        tn, fp, fn, tp = ub_test_cm.ravel()\n",
    "        ub_test_true_positive_rate = (tp / (tp + fn))\n",
    "        ub_test_false_positive_rate = (fp / (fp + tn))\n",
    "    else:\n",
    "        ub_test_true_positive_rate = np.nan\n",
    "        ub_test_false_positive_rate = np.nan\n",
    "    \n",
    "    sub_df = pd.DataFrame([[i, label_0, label_1, prediction_0, prediction_1, test_auroc2, test_balanced_accuracy, test_accuracy2, test_recall, test_precision, test_f1, test_auprc, test_cm, test_true_positive_rate, test_false_positive_rate, ub_prediction_0, ub_prediction_1, ub_test_auroc2, ub_test_balanced_accuracy, ub_test_accuracy2, ub_test_recall, ub_test_precision, ub_test_f1, ub_test_auprc, ub_test_cm, ub_test_true_positive_rate, ub_test_false_positive_rate]])\n",
    "    treatment_length_df = pd.concat([treatment_length_df, sub_df], axis=0, ignore_index=True)\n",
    "treatment_length_df.columns = ['iv_treatment_length', 'label_0', 'label_1', 'prediction_0', 'prediction_1', 'auroc', 'balanced_accuracy', 'accuracy', 'recall', 'precision', 'f1', 'auprc', 'cm', 'tpr', 'fpr', 'ub_prediction_0', 'ub_prediction_1','ub_auroc', 'ub_balanced_accuracy', 'ub_accuracy', 'ub_recall', 'ub_precision', 'ub_f1', 'ub_auprc', 'ub_cm', 'ub_tpr', 'ub_fpr']\n",
    "treatment_length_df.sort_values(by=['iv_treatment_length'], inplace=True)\n",
    "# Set to string\n",
    "treatment_length_df['iv_treatment_length']= treatment_length_df['iv_treatment_length'].astype(str)\n",
    "# Change 999\n",
    "treatment_length_df['iv_treatment_length'] = treatment_length_df['iv_treatment_length'].replace(['999'], 'PO')\n",
    "# Rename\n",
    "treatment_length_df.rename(columns={'iv_treatment_length': 'Proir days cumulative IV treatment length', 'label_0': 'Continue with IV', 'label_1': 'Switch to PO'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "df1 = treatment_length_df[['Proir days cumulative IV treatment length', 'Continue with IV', 'Switch to PO']].set_index(['Proir days cumulative IV treatment length'])\n",
    "df2 = treatment_length_df[['Proir days cumulative IV treatment length', 'prediction_0', 'prediction_1']].set_index(['Proir days cumulative IV treatment length'])\n",
    "df3 = treatment_length_df[['Proir days cumulative IV treatment length', 'ub_prediction_0', 'ub_prediction_1']].set_index(['Proir days cumulative IV treatment length'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Labels and predictions by IV treatment duration'}, xlabel='Proir days cumulative IV treatment length', ylabel='Count'>"
      ]
     },
     "execution_count": 191,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAhwAAAEWCAYAAAAtjU6HAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAABPfklEQVR4nO3deXhU5dn48e+dhCVsgUDYkkBQUFZBQaTggogWFRGrIBUUfWuxVYqtlaK+vlZ9q+IPsb5YoFpUoK5BW0FcUVlUrDRQ2XcJhD0gS4AQSOb+/XHOwBAmyWSZNffnuubKzHO2+5yZzLnnOc95HlFVjDHGGGOCKS7cARhjjDEm9lnCYYwxxpigs4TDGGOMMUFnCYcxxhhjgs4SDmOMMcYEnSUcxhhjjAk6SzhCSEQWiMjdoV62qoQzBhF5XERed5+3EpEjIhJfgfU8IiLTqj7CUrepItI2lNs0gQn1eyMiH4vIyFBtz5hIYglHBYhItoj0D3cc1ZWqblPVeqpaVNp8ItJXRLYXW/ZpVQ1r4lZRInKniHztPv9URJ70M8+NIrJbRBL8TAtqwugbXzhE2v+lb5LsparXquqMcMVkTDhZwmFCzt/J0JTbdOB2EZFi5bcDb6hqYXlXaO9L4OxYGVN+lnBUIRFpJCJzRSRXRA64z9OKzXauiCwRkUMiMltEkn2W7yUii0XkoIgsF5G+JWynrYgsdNexT0TeKSWmWe4v3kMiskhEOvlMmy4ik0XkQxHJE5HvRORcn+lXi8g6d9m/AMVPbr7beVxE3hWRd9x1LRORrj7Ts0VknIisAI6KSEJp+ysibdx9zBOReUATn2kZblV4gvs6WUReE5Gd7nF/X0TqAh8DLd3LL0dEpGXxX50iMkhEVrsxLBCRDsViflBEVrjH4B0Rqe1Oa+K+vwdF5EcR+UpESvt/uk5EfnDfrwkiEicitdxlu/hss6mI5ItISinrAngfSAYu81m2ETAQmOnn/XnKnfcv7rH4i1uuInKfiGwENrplA0Xke3ffFovIBT7reUhENrvvyxoRuckt7wD8FfiJu/6Dbvl0EZkizqWEIyLyjYg0F5EX3PdqnYhc6LP+liLynjj/Q1tEZIzPtMdFJFNEZrrbXy0iPdxpfwdaAR+42/mDv4MmImNFZJf7WfmvYtPOqAGSYjU2JRyr/xORHBE5LCJLReQyt3wA8AhwqxvP8uLbcD8Dj4rIVhHZ6+5XkjvN+xkfKSLb3M/Nf/vbJ2Oihqrao5wPIBvo76e8MXAzUAeoD8wC3veZvgDYAXQG6gLvAa+701KB/cB1OIng1e7rFJ9l73afvwX8tztfbeDSUmL9LzeWWsALwPc+06YDPwI9gQTgDeBtd1oT4DBwC1AD+B1Q6I3Bz3YeB076zP8gsAWo4XPMvgfSgcQA9vdb4Hk37suBPJ9jlQEokOC+/hB4B2jkbvsKt7wvsN1PnN71nAccdbddA/gDsAmo6RPzEqAlzsl9LfArd9ozOCfYGu7jMkBKODYKzHfX0QrY4PNeTgGe9Zn3fuCDEtZzJ/C1z+u/AdN8Xt/j+/76Wf7UZ6hYbPPc2BKBi4C9wCVAPDDSPQ613PmHuMcjDrjVPX4t/MXn8xnbB3TH+ax+6X4u7nDX/ydgvjtvHLAUeAyoCZwD/AD81Oe9O47zmYl334N/lfV/6TN9ALCH0/9/b7r739bf8fFzvM84Vm7ZCJz/+wTg98BuoHbxz5q/9wDnf3OTu5/1gH8Afy/2Gf+b+750BQqADuH+/rOHPSr6sBqOKqSq+1X1PVU9pqp5wFPAFcVm+7uqrlLVo8D/AEPFafw4AvhIVT9SVY+qzgOycL5cizsJtAZaqupxVS3xurmqvqqqeapagPMF2NX7K8r1D1Vdok4V/BtAN7f8OmCNqr6rqidxkpXdZRyCpT7zP49zgunlM32Squaoan5p+ysirYCLgf9R1QJVXQR84G+DItICuBYnETigqidVdWEZcXrdCnyoqvPcmJ/D+XLvXSzmnar6oxtDN7f8JNACaO1u8ytVLW1gomdV9UdV3YZzLH/uls8AbvOpHbkd+HuA8c8AhohIovv6DresvJ5xY8sHfgm8pKrfqWqROu0NCnDfR1Wd5R4Pj6q+g/NLv2cZ6/+nqi5V1ePAP4HjqjpTnTY47wDeGo6LcRLOJ1X1hKr+gHPCHeazrq/dz0wRznHqSuCGAq/5/P89Xo5lvXyPFar6uvt/X6iqE3ES5PMDXNdw4HlV/UFVjwAPA8PkzMs1T6hqvqouB5ZTvv01JqJYwlGFRKSOiLzkVpEeBhYBDeXMuylyfJ5vxfl13AQngRjiVmMfdKukL8U5qRX3B5zLG0vcauX/8jMPIhIvIuPdKvDDOL8AwefyBGcmEcdwfmmB8yv2VKzuydQ3dn985/cA2931nDWd0ve3JXDAPSl4bS1hm+nAj6p6oIzY/Gnpu1435hyc2hevko7PBJxfp5+5l0oeKmNbxd/3lu42v8OpJbhCRNoDbYE5gQTvJpq5wI0icg7OCfvNQJYtJbbWwO+LvS/p3nhF5A6fyy0HcWoLmhRfYTF7fJ7n+3ntPaatcS6B+W77EaCZz/zF34/aEnh7ijM+05T8mSrNGf8DIvJ7EVkrziW3g0ASZR8P33h8Y9iKU1NS2v7Ww5goZQ2fqtbvcX7dXKKqu0WkG/Afzmz7kO7zvBXOL+V9OF9kf1fVX5a1EVXdjfNLFBG5FPhcRBap6qZis94G3Aj0x0k2koADxeIpyS7fWEVEisXuj+/8cUAasNM3dJ/nJe6viLQGGolIXZ+ko1Wx5X3XkywiDVX1YLFpZQ2FvBPwbT/h3ccdZSyHW4P1e5yTcydgvoj8W1W/KGGRdGC1+7wVZx6XGTg1PruBd92agEDNxKnZOB/4TFX3lDJvScej+PvylKo+VXwm9335G3AV8K2qFonI95z+PFV26OkcYIuqtqvg8mVt/4zPNM774OsozuVQr+albcNtrzEO53isVlWPiPj+fwXy+WtdLJ5CnISseNsvY6Ke1XBUXA0Rqe3zSMBpK5EPHBSnMegf/Sw3QkQ6ikgd4EmcE0wR8Dpwg4j81K2ZqC3ObZ1nffGIyBCf8gM4X2z+bhGtj1Mdvh/ni/Tpcuzfh0AnEfmZu29j8P8F7Ku7z/y/dbf9rxLmLXF/VXUrzuWVJ0SkpptU3eBvJaq6C6dx6BRxGu3WEJHL3cl7gMbFLiH5ygSuF5GrRKQGTgJRACwuYz+9DSvbuknKYZzjX9ptumPd+NJx2mn4NvT9O3ATTtJxVoPPMszESSh/SdmXU/bgtBcozd+AX4nIJeKoKyLXi0h9nHYPilOrgojchVPD4bv+NBGpWc598FoCHBancXGi+7noLCIXB7h8WfuXCdzp8/9X/P/ze+Bnbk1lW+AXZWyvPk6CkAskiMhjQINi8WRIyY2J3wJ+J04D6Xo4/5/vaAXuMDImGljCUXEf4SQX3sfjONfmE3FqLP4FfOJnub/jNKTbjdPGYQyAqubg1EY8gvMFlgOMxf97dDHwnYgcwal+v19Vt/iZbyZONe0OYA0ln/zPoqr7cBoIjsdJWNoB35Sx2GycdhEHcNoi/MxtG+Fv/WXt7204DRd/xDkxlHYivh2npmgdToPH37rbWIfzpf6DW0Xve3kHVV2Pc5J/Eec9uwG4QVVPlLGf4ByPz4EjOA1cp6jqglLmn43TIPJ7nGTuFZ84tgPLcE7mXwWwbd99yMZJkOpS9qWY/wNuEefukEklrC8LJ3n5C877uAmn8SSqugaYiLO/e3Bqh3w/E1/i1OLsFpF95dkPd/1FOO9BN5yGpfuAaTg1c4F4BnjUfa8f9LP+j3H+R7/E2a8vi83yZ+AEzr7NwGnTVJpPcZLdDTj/Z8c585LLLPfvfhFZ5mf5V3G+Dxbh7O9x4DdlbNOYqCWlt3MzJjAi8jhOa/8R4Y4lGonIq8BOVX003LEYY0wwWBsOY8JMRDKAn3H6bg1jjIk5dknFmDASkf8FVgETSrgsZowxMcEuqRhjjDEm6KyGwxhjjDFBF7NtOJo0aaIZGRnhDsMYY6LK0qVL96lqWWP5GFNuQU84xOllMwvYoaoD3f4p3sEZKyAbGOrtJVJEHsa5970IGKOqn7rl3XFuJU3EuR31/jK6kSYjI4OsrKxg7JIxxsQsEalID6zGlCkUl1Tuxxn0yush4Au3N8Ev3NeISEecMRM64QyyNEVOdwk+FRiF0/dBO3e6McYYY6JEUBMOtzfM63E67/G6kdM9Is4ABvuUv+0O1rUFp2OenuIMztVAVb91azVm+ixjjDHGmCgQ7BqOF3AGGvP4lDVzu6P2dkvd1C1P5cxe+ra7Zanu8+LlZxGRUSKSJSJZubm5VbIDxhhjjKm8oCUcIjIQ2KuqSwNdxE+ZllJ+dqHqy6raQ1V7pKRYmydjjDEmUgSz0WgfYJCIXIczZkgDEXkd2CMiLVR1l3u5ZK87/3bOHMnRO9Lods4cObH4CKTGGGOMiXBBq+FQ1YdVNU1VM3Aag37pjrMxBxjpzjYSZ1Ar3PJhIlJLRNrgNA5d4l52yRORXu7InHf4LGOMMcaYKBCOfjjGA5ki8gtgG86IpKjqahHJxBnVtBC4zx09EuDXnL4t9mP3YYwxxpgoEbNdm/fo0UOtHw5jjCkfEVmqqj3CHYeJPda1uTHGGGOCLma7No8ajyeVe5EF2YUMmZXPrCGJ9M0I4C18/FAFAjPGGGOqjtVwRJnyJhsLsgtDEJUxxhhTOks4okhFko0hs/JDEJkxxhhTOks4okRFk41ZQxJDEJ0xxhhTOks4okBlko2A2ngYY4wxQWYJR4SzZMMYY0wssDNSmGUcf7PU6Udn3WDJhjHGmKhnNRwRzpINY4wxscASjghnyYYxxphYYAlHlLNkwxhjTDSwhCPKWbJhjDEmGljCEeUs2TDGGBMNLOGIcpZsGGOMiQaWcMQ4G0vFGGNMJLCEI4bZWCrGGGMihSUcMcrGUjHGGBNJgpZwiEhtEVkiIstFZLWIPOGWPy4iO0Tke/dxnc8yD4vIJhFZLyI/9SnvLiIr3WmTRESCFXcssFtljTHGRJpgno0KgH6qekREagBfi8jH7rQ/q+pzvjOLSEdgGNAJaAl8LiLnqWoRMBUYBfwL+AgYAHyMOYslG8YYYyJR0Go41HHEfVnDfWgpi9wIvK2qBaq6BdgE9BSRFkADVf1WVRWYCQwOVtzRzJINY4wxkSqobThEJF5Evgf2AvNU9Tt30mgRWSEir4pII7csFcjxWXy7W5bqPi9ebnxYsmGMMSaSBTXhUNUiVe0GpOHUVnTGuTxyLtAN2AVMdGf31y5DSyk/i4iMEpEsEcnKzc2tZPTRw5INY4wxkS4kd6mo6kFgATBAVfe4iYgH+BvQ051tO5Dus1gasNMtT/NT7m87L6tqD1XtkZKSUrU7EaEs2TDGGBMNgnmXSoqINHSfJwL9gXVumwyvm4BV7vM5wDARqSUibYB2wBJV3QXkiUgv9+6UO4DZwYo7mliyYYwxJloE8yzVApghIvE4iU2mqs4Vkb+LSDecyyLZwD0AqrpaRDKBNUAhcJ97hwrAr4HpQCLO3SnV/g4VSzaMMcZEk6CdqVR1BXChn/LbS1nmKeApP+VZQOcqDTCKWbJhjDEm2lhPo1GmvMmGjaVijDEmEljCEUUqkmzYWCrGGGMigSUcUaKiyYaNpWKMMSYSWMIRBSqTbFgbD2OMMZHAEo4IZ8mGMcaYWGAJR4SzZMMYY0wssIQjwlmyYYwxJhZYwhHhLNkwxhgTCyzhiHKWbBhjjIkGlnBEOUs2jDHGRANLOKKcJRvGGGOigSUcUc6SDWOMMdHAEo4YZ2OpGGOMiQSWcMQwG0vFGGNMpLCEI0bZWCrGGGMiiSUcMchulTXGGBNpLOGIMZZsGGOMiUSWcMQQSzaMMcZEqqAlHCJSW0SWiMhyEVktIk+45ckiMk9ENrp/G/ks87CIbBKR9SLyU5/y7iKy0p02SUQkWHFHK0s2jDHGRLJg1nAUAP1UtSvQDRggIr2Ah4AvVLUd8IX7GhHpCAwDOgEDgCkiEu+uayowCmjnPgYEMe6oY8mGMcaYSBe0hEMdR9yXNdyHAjcCM9zyGcBg9/mNwNuqWqCqW4BNQE8RaQE0UNVvVVWBmT7LVHuWbBhjjIkGQW3DISLxIvI9sBeYp6rfAc1UdReA+7epO3sqkOOz+Ha3LNV9Xrzc3/ZGiUiWiGTl5uZW6b5EIks2jDHGRIugJhyqWqSq3YA0nNqKzqXM7q9dhpZS7m97L6tqD1XtkZKSUu54o4klG8YYY6JJSO5SUdWDwAKcthd73MskuH/3urNtB9J9FksDdrrlaX7Kqy1LNowxxkSbYN6lkiIiDd3niUB/YB0wBxjpzjYSmO0+nwMME5FaItIGp3HoEveyS56I9HLvTrnDZ5lqp7zJho2lYowxJhIE8+dxC2CGe6dJHJCpqnNF5FsgU0R+AWwDhgCo6moRyQTWAIXAfapa5K7r18B0IBH42H1UOxVJNobMyid3evBjM8YYY0oTtIRDVVcAF/op3w9cVcIyTwFP+SnPAkpr/xHzKpps2FgqxhhjIoH1NBoFKpNsWBsPY4wxkcASjghnyYYxxphYYAlHhLNkwxhjTCywhCPCWbJhjDEmFljCEeEs2TDGGBMLLOGIcpZsGGOMiQaWcEQ5SzaMMcZEA0s4opwlG8YYY6KBJRxRzpINY4wx0cDOVlEu4/ibpU4/vm0Fu0MUizHGGFMSq+GIYce3rSD3/fHhDsMYY4yxhCNWeZONlMEPhTsUY4wxxhKOWOSbbNRudUG4wzHGGGOsDUessWTDGFNVli5d2jQhIWEazmjd9gM1+nmAVYWFhXd37959b6g3bglHDLFkwxhTlRISEqY1b968Q0pKyoG4uDgNdzymcjwej+Tm5nbcvXv3NGBQqLdvGWuMsGTDGBMEnVNSUg5bshEb4uLiNCUl5RBOjVXIWQ1HhCvrtlewZMMYEzRxlmzEFvf9DEtlg9VwRDlLNowxxkSDoNVwiEg6MBNojtNQ5WVV/T8ReRz4JZDrzvqIqn7kLvMw8AugCBijqp+65d2B6UAi8BFwv6pW+6zbkg1jTChlPPRh96pcX/b465eWNc+2bdsS7r333lbLly+vU7NmTU1LSyt48cUXcy644IKC8m7vySefbPq73/1uX/369T0AV1xxRdv33ntvS5MmTYoqEn9V8MYAMG3atOSHHnooF2Du3Ln1J06c2Gz+/PmbSlv+5ptvzhg4cOChlStXJhYUFMjkyZN3eKctXrw4ccSIEef88MMPq4O7F4EJZg1HIfB7Ve0A9ALuE5GO7rQ/q2o39+FNNjoCw4BOwABgiojEu/NPBUYB7dzHgCDGHRUs2TDGxDqPx8OgQYPaXn755Xk5OTmrNm/evPqZZ57ZsXPnzhoVWd9LL73U7MiRI6fOewsXLtwUzmTDN4b9+/fHv/LKK00rup6RI0funz17drJv2euvv5588803/1j5KKtG0BIOVd2lqsvc53nAWiC1lEVuBN5W1QJV3QJsAnqKSAuggap+69ZqzAQGByvuaGDJhjGmOpg7d279hIQE/cMf/uCtEad37975AwYMOOLxeLjnnnvS2rVr1+m8887r+Le//a2Rd5mePXueP2DAgHPatGnTadCgQW08Hg9/+tOfmu7du7fGFVdccd4ll1xyHkBqamqXXbt2Jaxfv77mOeec02nYsGGt27Zt26lPnz7tjhw5IgA9e/Y8f9GiRXUAdu3alZCamtoFoLCwkHvuuSetc+fOHc4777yOEyZMaFI8/kcffbTZn/70p6YAv/jFL9J79ep1HsDs2bPr33jjjW18Y/j973+flpOTU6t9+/Yd77nnnjSAo0ePxhffj5J07dq1oEGDBoVffvllXW/ZnDlzku+4447YTzh8iUgGcCHwnVs0WkRWiMirItLILUsFcnwW2+6WpbrPi5f7284oEckSkazc3Fx/s0S98iYbx7etCEFUxhhT9VasWJHYtWvXY/6mzZw5s+HKlSsT165du/qLL77Y8Nhjj6Vt3bq1BsDatWsTJ0+enLNp06bV27ZtqzVv3rx6jz766N6mTZueXLhw4YbvvvtuQ/H1bdu2rfaYMWP2btq0aXVSUlLRzJkzG5291dNeeOGFJklJSUWrVq1au3z58rUzZsxIWbduXU3fea688soj33zzTT2A77//vs7Ro0fjCwoKZNGiRfUuvfTSPN95J06cuD09Pb1g3bp1a1566aXtJe1HaTHdfPPNP77xxhvJAF988UXdhg0bFnbp0qXcl56CJegJh4jUA94Dfquqh3Euj5wLdAN2ARO9s/pZXEspP7tQ9WVV7aGqPVJSUiobesSpSLJhY6kYY2LRV199VX/o0KE/JiQkkJ6eXnjJJZcc+frrr+sAdOnS5ei55557Mj4+nk6dOh3bvHlzzbLWl5qaWtC7d+98gAsvvPBYdnZ2rdLm//zzzxtkZmY2bt++fccLL7yww4EDBxLWrFlT23eeSy+99NjKlSvrHjhwIK5WrVrao0ePI1999VWdb7/9tn6/fv2OlBVTefdj5MiRP3744YeNioqKeOONN5JvueWWiKndgCDfFisiNXCSjTdU9R8AqrrHZ/rfgLnuy+1Aus/iacBOtzzNT3m1UtFkw8ZSMcZEqy5duuS///77fmsaSrtvoFatWqcmxsfHU1hY6O+H6xlq1qzpu4zm5+fHASQkJGhRkdPM49ixY6fWo6oyceLEbTfffPPh0uJIS0srmDx5cpOePXse6dq1a/7nn39ef+vWrbUuvPDC42XFVN79aNu27cnU1NSCjz76qP5HH33U6Jtvvllb1jZCKWg1HCIiwCvAWlV93qe8hc9sNwGr3OdzgGEiUktE2uA0Dl2iqruAPBHp5a7zDmB2sOKORJVJNqyNhzEmWt1www15J06ckIkTJ55qH7Fw4cI6H374Yb0rrrgi7913300uLCxk586dCUuWLKl32WWXHS1tfXXr1i06dOhQuc576enpBUuWLKkL8MYbb5xKfq6++upDU6dOTSkoKBCAFStW1Dp8+PBZ6+7du/eRyZMnN+vbt29e//7982bMmJHSsWPHY3FxZ86alJRUdPTo0Uqfk4cMGfLj2LFj01u1alVw7rnnnqzs+qpSMGs4+gC3AytF5Hu37BHg5yLSDeeySDZwD4CqrhaRTGANzh0u96mqt/Xwrzl9W+zH7qNasGTDGBMpArmNtSrFxcUxZ86czffee2/6Cy+80NxbY/Diiy/mXHvttUcWL15cr0OHDp1ERJ944ontrVq1KlyxouR2ayNHjtx37bXXtmvatOlJf+04/HnooYf23Hrrree8/fbbjS+77LJTtRm/+93v9mVnZ9fq0qVLB1WV5OTkkx999NHm4stfccUVeZMmTWrer1+/ow0aNPDUqlVL+/Tpc9bllObNmxd17979SLt27Tr169fv0A033HAowMN0hjvuuOPAo48+mv7000/nlD13aEmsdmfRo0cPzcrKCncYZcp46MNSp+dMuq3SyUb2+OsrHacxpnoQkaWq2gNg+fLl2V27dt0X7phM1Vq+fHmTrl27ZoR6u9bTaISzmg1jjDGxwBKOCGfJhjHGmFhgCUeUs2TDGGNMNLCEI8pZsmGMMSYa2PD0Ua5Kko3Hkyodx4LsQobMymfWkET6Zvj5WD1eoQbXxhhjYoTVcES5SKjZKDPZMMYYU+3Z2SHGOWOpBO+2WEs2jKlGHk+q0uHpefxQmf16jBs3rvl7773XOC4uTuPi4pgyZcrWfv36ldrBF8Bvf/vbln379s0bPHhwXvFh6f154IEHWtarV6/oySef3FPSPAAPPfRQ8/Hjx+8ua/u+UlNTu9StW7dIRGjSpMnJt956a0urVq0K9+/fH3/33XenZ2Vl1QPo0aPHkWnTpuU0btw4rCPYBktANRwi0ieQMhNZgj2WiiUbxphg+vzzz+t++umnDVeuXLlmw4YNa+bPn7/hnHPOORHIsi+88MLOwYMH58HZw9JXxqRJk1qUPdfZFi5cuGH9+vVrLrzwwmOPPfZYC4Dhw4e3btOmzYmcnJxVOTk5qzIyMk6MGDGidVXEGYkCfQNeDLDMRIhgj6ViyYYxJth27NhRIzk5uTAxMVEBWrRoUZiRkXFy/vz5da655ppzAV5//fWGtWvXvuj48eNy7NgxSUtL6wJw8803Z7z22muN/A1L/+677zbo2LFjh/PPP7/jT37yk/O821u7dm1iz549z09LS+viHVbe17333ptaUFAQ1759+46DBg1qA/D44483a9euXad27dp1evLJJ89apri+ffvmbdmypdaqVatqrVy5su7/+3//79TYYBMmTNi5YsWKuqtXry514LhoVeqZQkR+AvQGUkTkAZ9JDYD4YAZmKi7Yt8qWN9lYkF1I3yqPwhgT6wYPHnz4mWeeaZmRkdH50ksvPfzzn//8x+uvv/7IpZdeemz16tV1ABYtWlSvbdu2+YsWLapz8uRJufDCC8/oNvzRRx/dO3Xq1GYLFy7c0KJFi8KdO3cmjB49OmPBggXr2rdvf2LPnj2nzmWbNm2qvXjx4vUHDx6M79ChQ+exY8fm+g6gNmXKlB3Tp09vum7dujUAX331VZ0333yz8dKlS9eqKt27d+9w1VVX5fXp0ye/pH2aM2dOw44dO+YvX768dseOHY8lJJz+Dk1ISKBjx47Hvv/++9qdOnWKmGHlq0pZNRw1gXo4iUl9n8dh4JbghmYqIhKTjSGzSvzfM8aYEiUlJXlWrVq15i9/+cvWlJSUwpEjR547adKkxjVq1KB169bHly1bVnvZsmV1f/Ob3+yZP39+/YULF9b3N06JrwULFtTt2bNnXvv27U8ANGvW7FR7iWuuueZgYmKitmjRojA5Ofnk9u3bS/2SW7BgQb3rrrvuYIMGDTxJSUme66+//sD8+fPr+5v3iiuuOK99+/Yd8/Ly4v73f/93t6q645GeSVXxVx4LSj2YqroQWCgi01V1a4hiMhUUqcnGrCGJVR6LMaZ6SEhIYODAgXkDBw7Mu+CCC/L//ve/Nx4zZsz+3r17H5kzZ05SjRo19IYbbjh82223ZRQVFcnzzz9f6qBlpZ3QyzscfHnGIvPWsHhfd+vWLX/cuHF1ioqKiI93KlmKiopYu3ZtnQsuuKDMoeujUaBtOGqJyMsi8pmIfOl9BDUyUy6RnGxYGw9jTEUsX7681sqVK0+1Z/jPf/6TmJaWdgKgb9++R1566aWmF1988ZGWLVsWHjhwIOGHH36o3b1797NO1r7D0l955ZVHv/vuu/rr1q2rCeB7SSUQCQkJ6h2Svl+/fkc++uijhnl5eXGHDx+O++ijjxpdeeWVeYGsp3PnzgWdOnU6Nm7cuFONUMeNG9eic+fOxzp37hxzl1Mg8NtiZwF/BaYBMXm7TjSzZMMYExIB3MZalQ4fPhw/ZsyYVocPH46Pj4/XjIyMghkzZmwFJ+HYv39/jb59+x4B6NixY/6ePXsK4+LO/h1dfFj6SZMmZd90001tPR4PjRs3Prl48eKNgcY0fPjw3A4dOnTs3LnzsTlz5my57bbb9l900UUdAG6//fbc0tpvFPfGG29k33333a1atWrVWVW56KKLjr7xxhvZgS4fbQIant4drrhq778OslgZnr4sgSQbZQ5PX0ZPoykT8iqfbFhPo8ZEBRuePvZF+vD0H4jIvSLSQkSSvY+gRmbKFKqB26xmwxhjTGUFelYY6f4d61OmwDlVG44JVChHibVkwxhjTGUFdGZQ1TbBDsQELhKHpLdkw5iY5PF4PBIXFxf47Rgmonk8HgFK7OI9mALt2vwOf48ylkkXkfkislZEVovI/W55sojME5GN7t9GPss8LCKbRGS9iPzUp7y7iKx0p03ye/NyNVHeZMMZSyX4LNkwJiatys3NTXJPUibKeTweyc3NTQJWhWP7gZ4hLvZ5Xhu4ClgGzCxlmULg96q6TETqA0tFZB5wJ/CFqo4XkYeAh4BxItIRGAZ0AloCn4vIeapaBEwFRgH/Aj4CBgAfBxh7zKhIsuGMpfJw0GOzZMOY2FNYWHj37t27p+3evbszNrp4LPAAqwoLC+8Ox8YDvaTyG9/XIpIE/L2MZXYBu9zneSKyFkgFboRTPV3PABYA49zyt1W1ANgiIpuAniKSDTRQ1W/dbc8EBlPNEo6KJhvBGkvFGBP7unfvvhcYFO44TGyoaMZ6DGgX6MwikgFcCHwHNHOTEW9S4h3sJhXw7SFuu1uW6j4vXu5vO6NEJEtEsnJzcwMNL+JVJtmIhDYeC7ILy57JGGNMTAuohkNEPsC5KwWcQds6AJkBLlsPeA/4raoeLqX5hb8JWkr52YWqLwMvg9MPRyDxRbpYSDaGzMond3q4IzHGGBNOgV54f87neSGwVVW3lzSzl4jUwEk23lDVf7jFe0SkharuEpEWwF63fDuQ7rN4GrDTLU/zU14thCLZyDj+ZqnTs2vfFvC6fNlYKsYYY7wCuqTiDuK2Dmek2EbAibKWce8keQVYq6rP+0yaw+l+PUYCs33Kh4lILRFpg3PJZol72SVPRHq567zDZ5mYF+01G3b3ijHGGAj8ttihwBJgCDAU+E5Eyhqevg9wO9BPRL53H9cB44GrRWQjcLX7GlVdjXOZZg3wCXCfe4cKwK9xxnHZBGymGjUYtWTDGGNMLAj0bPDfwMWquhdARFKAz4F3S1pAVb/Gf/sLcG6r9bfMU8BTfsqzgM4BxlqtWLJhjDEmGgR6l0qcN9lw7S/HsiaILNkwxhgTDQI9K3wiIp8Cb7mvb8XpgMuEmSUbxhhjokGpZwYRaYvTb8ZYEfkZcCnOZZJvgTdCEJ8pgyUbxhhjokFZl0VeAPIAVPUfqvqAqv4Op3bjheCGZqpCsMdSsWTDGGNMIMpKODJU9awzltuIMyMoEZkqc3osleCwZMMYY0ygyko4apcyzXpzimDBHkvFkg1jjDHlUVbC8W8R+WXxQhH5BbA0OCGZygr2rbLlTTZsLBVjjDFlnS1+C/xTRIZzOsHoAdQEbgpiXKaCIjHZsLFUjDHGlHrGUNU9QG8RuZLTHW99qKpfBj0yU26RmmzYWCrGGGMCuviuqvOB+UGOxVRCJCcb1sbDGGOM9RYaAyzZMMYYE+ks4YhyoRhLxZINY4wxlWUJRxQL1cBtlmwYY4ypLEs4olQoR4m1ZMMYY0xlWcIRhSJtSHoo32UXY4wx1Y8lHFGmvMlGsMdS8bJkwxhjTGks4YgiFUk2gjmWii9LNowxxpQmaAmHiLwqIntFZJVP2eMiskNEvncf1/lMe1hENonIehH5qU95dxFZ6U6bJCISrJgjWUWTjWCNpWKMMcaURzBrOKYDA/yU/1lVu7mPjwBEpCMwDOjkLjNFROLd+acCo4B27sPfOmNaZZKNSGjjYWOpGGOMCVrCoaqLgB8DnP1G4G1VLVDVLcAmoKeItAAaqOq3qqrATGBwUAKOULGQbAyZlR/uMIwxxoRZONpwjBaRFe4ll0ZuWSqQ4zPPdrcs1X1evLzaiIVkw8ZSMcYYE+qEYypwLtAN2AVMdMv9tcvQUsr9EpFRIpIlIlm5ubmVDDUyxEKyYQ1KjTHGhDThUNU9qlqkqh7gb0BPd9J2IN1n1jRgp1ue5qe8pPW/rKo9VLVHSkpK1QYfJpZsGGOMiQUhTTjcNhleNwHeO1jmAMNEpJaItMFpHLpEVXcBeSLSy7075Q5gdihjjnSWbBhjjIkGQTsjiMhbQF+giYhsB/4I9BWRbjiXRbKBewBUdbWIZAJrgELgPlUtclf1a5w7XhKBj92HcVmyYYwxJhoE7aygqj/3U/xKKfM/BTzlpzwL6FyFocUUSzaMMcZEA+tpNMpZsmGMMSYaWMIR44I9loolG8YYYwJhCUcMC/ZYKpZsGGOMCZQlHDEq2GOpWLJhjDGmPCzhiEHBvlW2vMmGjaVijDHGfprGmEhMNobMyid3epWHYowJsqVLlzZNSEiYhnOnoP1ANaXxAKsKCwvv7t69+15/M1jCEUMiNdmwsVSMiU4JCQnTmjdv3iElJeVAXFxcicNKGOPxeCQ3N7fj7t27pwGD/M1jCUeMiORkI+A2Ho8nVTLKsuPJPeoJ2jaMCapK/H9Uos1VZ0s2TCDi4uI0JSXl0O7du0vsN8uqyGJATCQbQWQ1LaY6q+T/Y5wlGyZQ7melxLzCEo4oF4qxVGIl2YiEeIwJJWvgbSKJfQNHsVAN3GbJhjHRp6I1k6V54oknuldVfAB//OMfl5Y1T506dS48duzYf/xNu+uuu9I//PDDRrt27VoRHx8PwKRJkxr/8Y9/TGvWrNnJwsJCadu2bf6sWbOy69ev73nggQdavv76602Sk5NPZVZff/31+n/96191Jk6c2Gz+/PmbfNf/1ltvJT355JOpHo+HwsJC+dWvfrVn7Nix+yq529WW1XBEqVCOEmvJhjHRpTo08C4qKuKTTz5p2KJFixMff/xxfd9pN9xww4F169at2bRp0+oaNWroq6++2sg77Ve/+tWedevWrfE+mjRpUnT22qGgoEDuv//+1nPnzt24fv36NatWrVpzzTXX5AV7v2KZfRNHoUgbkh7gqswEUgY/xZ3NL4DjZ0/PDmEslmyY6iza21wFau7cufXPO++8/FtuueXAm2++mTxw4MCzkoGTJ09y7NixuOTkZL9JRWkOHjwYV1hYKM2aNSsESExM1K5duxZURezVldVwRJnyJhvBHkvFK1KSn2j98jSmKlSXZAPgzTffTB46dOiPw4cPP/D5558nFRQUiHfaBx980Kh9+/Ydmzdv3vXgwYMJP//5zw96p/31r39t1r59+47t27fveMkll5xX0vqbNWtWdPXVVx9s1arVBTfccEObqVOnJhcVlTtvMT4s4YgiFUk2gjmWii9LNowJv+qSbBw/flzmz5+fdNtttx1MTk72dOvW7eg///nPBt7p3ksqubm5yzt06JD/2GOPNfdO872k8t13320obTvvvPPO1k8++WRDjx49jk6aNKn50KFDM4K4WzEvuj5l1VhFk41gjaUSaaL5y9OYqlIdkg2A9957r0FeXl58586dOwHk5+fHJSYmeoYNG3bId764uDgGDRp0cPLkyU0ruq2ePXvm9+zZM3/UqFE/tm3btguhvUIcU6yGIwpUJtmIhJqHYF/WifYvT2OqSnVINgDefvvt5BdeeGHrjh07Vu7YsWNldnb2yq+++qpBXl7eWee0r776qn5GRka5214cOnQobu7cuacao3733XeJLVu2PFHZ2Kuz6Py0VSOxkGw4l3UeDsr6Y+HL05hQqez/SyC3sVa148ePxzVr1uzUl9ldd921d9GiRUkzZszY6i1r0KCBp0ePHkfefvvtJDjVhqOex+OhRYsWJ958881s77x//etfm2VmZjb2vp49e/YmgG+//baB73ZmzJjxw4QJE5qNHj26de3atT116tTxvPLKK1uCvLsxLWjf0CLyKjAQ2Kuqnd2yZOAdIAOnWmqoqh5wpz0M/AIoAsao6qdueXdgOpAIfATcr6rVpue7WEg2gnVZpyIN5PoGJRJjokM0Jucej+esJGf8+PG7i5d99tlnm73Px4wZs9/fup5//vmdzz///M7i5eeff/6J48ePLytePmDAgE3Fy0zFBfNTNx34CzDTp+wh4AtVHS8iD7mvx4lIR2AY0AloCXwuIuepahEwFRgF/Asn4RgAfBzEuCNKJCQbGcffLPcykdrduo1aa6qzaEs2TGwJ2idPVReJSEax4hvh1I/MGcACYJxb/raqFgBbRGQT0FNEsoEGqvotgIjMBAZTBQlHxkMfljo9u/ZtJU4L+GT3+KGSpwUo3MlGRURqshFNnRoZEwyWbJhwCnWj0WaqugvA/ettOZwK5PjMt90tS3WfFy/3S0RGiUiWiGTl5uZWaeBekdZmwJKNqp3fmOrMxlIxwRQpd6mInzItpdwvVX1ZVXuoao+UlJQqC84rEk9elmxU3fzGVGeBjKViTGWEOuHYIyItANy/e93y7UC6z3xpwE63PM1PechF6qiL1SXZgOrTqZExoWaXHU0ohDrhmAOMdJ+PBGb7lA8TkVoi0gZoByxxL7vkiUgvERHgDp9lQiYYoy5WleqSbED16dTImFCy/xcTKsG8LfYtnAaiTURkO/BHYDyQKSK/ALYBQwBUdbWIZAJrgELgPvcOFYBfc/q22I8J8R0q0d5A0el06/qgrt9GrTUmOpX3/6W8w9Nv2bKFWbNmMWTIENq0aXPW9ED69YiPj+/erl27/KKiImnbtm1+ZmZmdv369T2+5enp6QWZmZlbmjRpUrR+/fqaXbt27ZyRkXFqGMnRo0fvGT169P7U1NQudevWLQIoKiqS66+//sCzzz67KzExUdevX19z4MCB7TZu3LgaYP78+XXGjh2bvm/fvhoioj179jzSrVu3YzNmzEgB2Lx5c+02bdocj4uLo1+/fofat29/PCsrq+7MmTO3ATz33HNNJk+e3AygXr16nueeey7npz/96RGAnj17nn/s2LG4VatWrQVYtGhRnQcffDB9yZIl6/Py8uKGDx/eet26dYmqKg0aNCj88ssvNyYlJXnKc+wjUdBqOFT156raQlVrqGqaqr6iqvtV9SpVbef+/dFn/qdU9VxVPV9VP/Ypz1LVzu600aHsgyPa2wwEeyyVSGuwGmnH35hIFuz/l7KSjUDVqlXLs27dujUbN25cXaNGDZ04cWJK8fKGDRsWTpgw4VTDvfT09ALfIehHjx59ql+OhQsXbtiwYcOaZcuWrd2yZUut4cOHty6+zZycnIThw4efO378+O3Z2dmrNm/evHrAgAGHR4wYccC7zqZNm55cuHDhhnXr1q2ZMmXKDt/l33rrraTXXnstZfHixeu3bNmyeurUqVvvvPPONtu2bTt1oPfv35+QmZnZoPi2n3766aZNmzY9uWHDhjUbN25c/eqrr2bXrFkzJvqeipRGoxEnVpKNYHW6FWnJBkRnp0bGhEO0JBvFXXrppUc2bdpUq3h5r169ju7YsaNmedaVlJTkmTFjxtZ58+Y13LNnT7zvtIkTJzYdOnTo/v79+x8FZ0yWu+6660B6enpAjfOee+655s8888z2Fi1aFLpxHxs6dOj+iRMnnhrTZfTo0XvGjx/fsviyu3btqpGamnrS+7pr164FiYmJlnDEslhJNoKRDFSku/VQiJTjb0wki9Zk4+TJk3z66acNunTpckYDucLCQubPn19/8ODBB71lOTk5tbxD0Ldv377jJ598Us/fOpOTkz2pqaknVq9eXdu3fM2aNYk9evQ4VtFYN23alNinT58zlr/44ouPrV279tS19ssuu+xIzZo1PR988EF93/lGjRq178UXX2zerVu39mPGjGm5cuXKsxKsaGXfziWwZKNq1l+esVTK6tG0tM7YwDo1MqYs0ZhsFBQUxLVv374jwCWXXJJ3//337/Mt37FjR83OnTsfGzx48GHvMt5LKoGsP1RX6VUV596H0x555JFdTz/9dItnn332VH9TvXv3zt+yZcvK999/v8G8efMa9O7du8PChQvXXXTRRcfPWmmUsRqOEliyUfn1B/uyTnlZp0amOovGZANOt9VYt27dmhkzZuTUrl1bfcuzs7NXnjhxQsaPH1/uIegPHDgQt3PnzppdunQ542TeoUOH/KysrDoVjblt27b533zzzRnLL126tE779u3PqJ0ZNGhQXkFBQdzXX39d17c8KSnJM3LkyIOvv/76tptuuunH2bNnJ1U0lkhiCUclWLIRvnjKyzo1MtVZtCYbgWjcuHHRpEmTtk2ePLlZQUGBv84i/Tp06FDcXXfd1frqq68+mJKSUuQ77cEHH9ybmZnZ+MsvvzyVCEyZMiXZt9FnaR544IHdjzzySNru3bvjARYvXpz4zjvvNH7ggQfO6gJ73Lhxu1588cXm3tefffZZ3dzc3HiA48ePy4YNG2pnZGScCHS/Iln4z5RRLP7Gp7iz+QVQQkVXdojisGSjdJF2q7IxoVSVyUZpt7FOnz69UuuujD59+uR36NAhf9q0aY369+9/xNuGwzt9xIgR+x599NG9AFdcccV5qioej4frrrvu4LPPPntWZ5Lp6emFM2fO/GHs2LFp+/fvrxEXF6e9evU6cvvttx8MJJ7hw4cf2r59e81evXp1EBGtW7eu59VXX93SunXrk8XnvfXWWw89+eSTp6pfN2zYUHv06NGtATwej/Tv3//QyJEjD1TgsEQcidWR3nv06KFZWVklTi9r8LaqkD2+7P4vKhtHICf3suIoK4acSbdVOtmoimNRVhsOf876sq2CAfWMCYvHy1+rXpG77a6ccWypqvYAWL58eXbXrl33lT9YU10tX768SdeuXTP8TbNLKlEsVDUJsVCzEQmXvYwJpUjuIdlUT5ZwBEmwbwUN5cndkg1joku095BsYpMlHEFQ3Xr4hMgatdaSDVOdRXunhSZ2WcJRxSKth89QdbplyYYx4WfJholklnBUoUi9WyQULNkwJvws2TCRzBKOKhKpyUakdLoVbPblaUz09pBsqgf7pFWBSE42IqHmwbmsU/ZtsRVlX57GOIKdbAQyPH15OgELZHj6cePGNX/vvfcax8XFaVxcHFOmTNn6zDPPNM/Jyal17NixuAMHDiSkpqaeAHjxxRe3/vd//3fa3r17a9SqVctTo0YNffnll7N79+6dD5CamtolKytrbYsWLQq3bduWcO+997Zavnx5nZo1a2paWlrB+PHjd9x1111tAHbt2lWzXr16RfXr1y9KTk4uXLx48QbfuOLj47u3a9cuv6ioSNq2bZufmZmZXb9+fY9veXp6ekFmZuaWJk2aFK1fv75m165dO2dkZJzquWn06NF7Ro8evT81NbVL3bp1iwCKiork+uuvP/Dss8/uSkxM1PXr19ccOHBgu40bN64GmD9/fp2xY8em79u3r4aIaM+ePY9069bt2IwZM1IANm/eXLtNmzbH4+Li6Nev36H27dsfz8rKqjtz5sxtAM8991yTyZMnNwOoV6+e57nnnsv56U9/egSgZ8+e5x87dixu1apVawEWLVpU58EHH0xfsmTJ+ry8vLjhw4e3XrduXaKqSoMGDQq//PLLjUlJSZ6y3kMvq+GoJEs2AosnWCzZMCZw0dbj6Oeff173008/bbhy5co1GzZsWDN//vwN55xzzol58+ZtdoeF39qjR48j3q7Pr7766qMAM2fO/GH9+vVrfvnLX+598MEH04qv1+PxMGjQoLaXX355Xk5OzqrNmzevfuaZZ3YcOnQo3ruu/v37H/zTn/60fd26dWuKJxtwumv1jRs3rq5Ro4ZOnDgxpXh5w4YNCydMmJDiXcY7xov3MXr06P3eaQsXLtywYcOGNcuWLVu7ZcuWWsOHD29dfJs5OTkJw4cPP3f8+PHbs7OzV23evHn1gAEDDo8YMeKAd51NmzY9uXDhwg3u8dnhu/xbb72V9Nprr6UsXrx4/ZYtW1ZPnTp165133tnGtwfV/fv3J2RmZjYovu2nn366adOmTU9u2LBhzcaNG1e/+uqr2TVr1ixXR16WcFRCKE7usZBsBOuyTkUayBlTnUVSsrFly5Yy59mxY0eN5OTkQu/w7C1atCjMyMg4q7fOklx++eVH9+zZc9aw9XPnzq2fkJCgf/jDH051Nd67d+/8AQMGHAl03b4uvfTSI5s2bTprVNdevXod3bFjx1nbL01SUpJnxowZW+fNm9dwz5498b7TJk6c2HTo0KH7+/fvfxQgLi6Ou+6660B6enpAX27PPfdc82eeeWZ7ixYtCt24jw0dOnT/xIkTT41DM3r06D3jx49vWXzZXbt21UhNTT117Lt27VrgfV8CZQlHBVmnW+GNxzo1Mqb8IinZmDVrVpnzDR48+PDOnTtrZmRkdB4xYkSrDz/80O8w8yX54IMPGlx77bUHi5evWLEisWvXrhUeft7XyZMn+fTTTxt06dLljC+YwsJC5s+fX3/w4MGntu/tct37+OSTT/zuT3Jysic1NfXE6tWra/uWr1mzJrFHjx4VjnvTpk2Jffr0OWP5iy+++NjatWtPdcBy2WWXHalZs6bngw8+qO8736hRo/a9+OKLzbt169Z+zJgxLVeuXHlWglUWq4OuAOt0K7zxBLVTowp0H11u1r26CZNISjaGDBlS5rxJSUmeVatWrfnkk0/qf/HFF/VHjhx57mOPPbZ9zJgx+0tb7o477jgnPz8/zuPxkJWVtTbwvQlcQUFBnHe8lksuuSTv/vvv3+dbvmPHjpqdO3c+Nnjw4MPeZbyXVAJZf6iGHVFVRM4c8+6RRx7Z9fTTT7d49tlnt3vLevfunb9ly5aV77//foN58+Y16N27d4eFCxeuu+iii0oYTexsYanhEJFsEVkpIt+LSJZbliwi80Rko/u3kc/8D4vIJhFZLyI/DUfMXtXt5B5p8UR7PwN2WcdEsvJ+PiuTbATaxiMhIYGBAwfm/fnPf945YcKEbe+//36jspaZOXPmD9u2bVs5ePDgH3/5y1+2Kj69S5cu+cuXL6/w8PNwuq3GunXr1syYMSOndu3a6luenZ298sSJEzJ+/PimZa2ruAMHDsTt3LmzZpcuXc44mXfo0CE/KyurwnG3bds2/5tvvjlj+aVLl9Zp3779GbUzgwYNyisoKIj7+uuv6/qWJyUleUaOHHnw9ddf33bTTTf9OHv27HL9QgvnJZUrVbWbd5Ag4CHgC1VtB3zhvkZEOgLDgE7AAGCKiMT7W2GwRdrJHSKrh89wJxsZx98849F8w0NclZngjuqbScbxN6s8pvKwyzomkpX38xmKZGP58uW1fKvu//Of/ySmpaUFNFR7rVq19M9//vOO77//vu6yZcvOuDRxww035J04cUImTpzYxFu2cOHCOuW9ZFOaxo0bF02aNGnb5MmTmxUUFEjZSzgOHToUd9ddd7W++uqrD6akpBT5TnvwwQf3ZmZmNv7yyy9PJQJTpkxJ9m30WZoHHnhg9yOPPJK2e/fueIDFixcnvvPOO40feOCB3OLzjhs3bteLL77Y3Pv6s88+q5ubmxsPcPz4cdmwYUPtjIyMgN4Lr/D/5DvtRqCv+3wGsAAY55a/raoFwBYR2QT0BL4NZXAV6+EzeLeCelWXZAPK1+At0pJDG6vCRDLfz+eVM0puIlD8NtbyDklf3vkPHz4cP2bMmFaHDx+Oj4+P14yMjIIZM2ZsDXT5evXq6a9//es948ePb5aZmXlqubi4OObMmbP53nvvTX/hhRea16pVS9PS0gpefPHFnHIFWIY+ffrkd+jQIX/atGmN+vfvf8TbhsM7fcSIEfseffTRvQBXXHHFeaoqHo+H66677uCzzz67s/j60tPTC2fOnPnD2LFj0/bv318jLi5Oe/XqdeT2228/GEg8w4cPP7R9+/aavXr16iAiWrduXc+rr766pXXr1mc1xL311lsPPfnkk6eqvDZs2FB79OjRrQE8Ho/079//0MiRIw+U53iEZXh6EdkCHAAUeElVXxaRg6ra0GeeA6raSET+AvxLVV93y18BPlbVd/2sdxQwCqBVq1bdt24t+XNZnmHhK3pratGxsq/VV3Z4+kBUdnj6sgRyfKpiePrpu4eWmmx4azBKi6eyxyK79m2lTvfnrJoZa8NhKiJI7YuKfz7licM2PL2psEgcnr6Pql4EXAvcJyKXlzKvv6oov1mSqr6sqj1UtUdKSoq/Wcot2nv4jKVRa6O9ZiMS2pAY48s+nyaUwpJwqOpO9+9e4J84l0j2iEgLAPfvXnf27UC6z+JpwFlVTcFgnW4Ftv5I299Iice+zE0ks8+nCbWQf8pEpC4Qp6p57vNrgCeBOcBIYLz7d7a7yBzgTRF5HmgJtAOWBDvOWEk2ImXU2qpQVqPP3Pdvi5jjb1/mJpKV4/Pp8Xg8EhcXF/pr7ybqeDweAUrs6jwc34TNgH+69/0mAG+q6ici8m8gU0R+AWwDhgCo6moRyQTWAIXAfapa5H/VVSdWko1I6W69OjWgtWTDRLJyfj5X5ebmdkxJSTlkSYcpjcfjkdzc3CRgVUnzhPzbUFV/ALr6Kd8PXFXCMk8BTwU5tDNYslE16z99WefhKo+luEg4/pZsmEhW3s9nYWHh3bt37562e/fuzljP1KZ0HmBVYWHh3SXNYN+IJbBko/Lrj8wGtDZqrameKvL57N69+15gUHAjM9WFZayVYMlG+OIpLxu11lRn9vk0kcA+eZUQaSdTSzbKjicYqvzLvAr6W0iZkFd2PNYfSLVgyYaJFFbDUQmRdjINVjyxkmxEytguoWAnFwOR+/k01ZN9I0WxUJ3cLdnwr6IDyeVOr/JQzmLJhvH3+Szt1vJgX3Y0xmo4giSWevi0ZONslRm1NhLYL9nYVt7PZ6Q18DaxyX4GBUGwbwWNtJM7VK8GtJVJNgKZP5BRbSsypkvxeEJR02JCrzLJRiT8/5rYZTUcVSzSevgMdk2LV6R8WUV7shFskVbTYqqWJRsmkoX/GzCGROrdItWl061QfHkG8mV+9qi1T3Fn8wvguDM9OyiRlS3Skh9T9a7KTDjr81YSSzZMqFkNRxWJ1GSjulyTDdWXZ7T+crRko3qI1gbepnqwhKMKRHKyEQlfJrHUgNaSDRPJovH7wVQf9u1TSZZsBBZPrDSgDaRBZ8yMWhtAB2SVbtNinY+FVKR9P5jqxWo4KsE63Qo8nmCv3xrQni3YNRvR3oC2Ooqk7wdT/dh/fQVZp1vhjcca0JauKk7uZdXmHJ11gyUbUSZSvh9M9WT/+RVgnW6FNx5rQFu6BdmFAd2tkF3J7ViyEX0i4fvBVF/2319O1e3kHmnxxEKbFri+0uspqfYh0hrQWrIRXUJ12dFUT/YNUA6RdvKCyLoma8lGYPHESgPaQFQ62aiCkXO9Sk1+rPGqjaVigs4SjgBVrIFi5X/JliVSTi7WgDbweIK9/kj6fMbfWLnLOlXVzbvVtJTO9/Oz561Hwh2OiVFR858nIgOA/wPigWmqGrJU3Bools4a0IY3nnB9PgNJBmq3qtQmqoQlG6WLtP8XE7ui4r9PROKBycDVwHbg3yIyR1XXBHvb0d5AMdi/ZK0BbXjjifbPZ7CFqgFttIq0/xcT26Ii4QB6AptU9QcAEXkbuBEIasJhbQYCW3+k7W91iScWPp9R04C2CtqSlFnTEuJ2JJH2eTCxT1Q13DGUSURuAQao6t3u69uBS1R1dLH5RgGj3JfnA+urOJQmwL4qXmc0xgCREUckxACREUckxACREYfFcFpF4mitqinBCMZUb9FSwyF+ys7KlFT1ZeDloAUhkqWqPYK1/miJIVLiiIQYIiWOSIghUuKwGCIvDmMgero23w6k+7xOA3aGKRZjjDHGlFO0JBz/BtqJSBsRqQkMA+aEOSZjjDHGBCgqLqmoaqGIjAY+xbkt9lVVXR2GUIJ2uaYcIiEGiIw4IiEGiIw4IiEGiIw4LIbTIiUOY6Kj0agxxhhjolu0XFIxxhhjTBSzhMMYY4wxQWcJRwBEZICIrBeRTSISli4aReRVEdkrIqvCsX03hnQRmS8ia0VktYjcH6Y4aovIEhFZ7sbxRDjicGOJF5H/iMjcMMaQLSIrReR7EckKUwwNReRdEVnnfj5+EoYYznePgfdxWER+G4Y4fud+LleJyFsiUjsE2yxy93mViMwSkTpueZqIzBaRjSKyWUT+z214b0zIWcJRBp9u1a8FOgI/F5GOYQhlOjAgDNv1VQj8XlU7AL2A+8J0LAqAfqraFegGDBCRXmGIA+B+YG2Ytu3rSlXtFsY+F/4P+ERV2wNdCcMxUdX17jHoBnQHjgH/DGUMIpIKjAF6qGpnnEbuw0Kw6Xx33zsDJ4BfiYgA/wDeV9V2wHlAPeCpEMRjzFks4SjbqW7VVfUE4O1WPaRUdRHwY6i3WyyGXaq6zH2eh3NSSQ1DHKqqR9yXNdxHyFs/i0gaTt/c00K97UgiIg2Ay4FXAFT1hKoeDGtQcBWwWVW3hmHbCUCiiCQAdQh9n0FfAW2BfsBxVX0NQFWLgN8B/+WtATEmlCzhKFsqkOPzejthOMlGGhHJAC4EvgvT9uNF5HtgLzBPVcMRxwvAHwBPGLbtS4HPRGSp271/qJ0D5AKvuZeXpolI3TDE4WsY8FaoN6qqO4DngG3ALuCQqn4Wqu27Sc61wEqgE7C0WHyH3djahiomY7ws4ShbQN2qVyciUg94D/it+wUWcqpa5FadpwE9RaRzKLcvIgOBvaq6tMyZg6+Pql6Ec6K5T0QuD/H2E4CLgKmqeiFwFAjbcLRuG4VBwKwwbLsRTg1oG6AlUFdERoRg04luAp6Fk1C8gvPd5e+7qqRyY4LKEo6yWbfqPkSkBk6y8Yaq/iPc8bhV9wsIffuWPsAgEcnGuczWT0ReD3EMAKjqTvfvXpw2Cz1DHMJ2YLtPLdO7OAlIuFwLLFPVPWHYdn9gi6rmqupJnDYUvUOwXW8bjm6q+hv38u9q4Iw2Pe7lr3RgcwhiMuYMlnCUzbpVd7mN0F4B1qrq82GMI0VEGrrPE3G+5NeFMgZVfVhV01Q1A+cz8aWqhuKX7BlEpK6I1Pc+B64BQnonk6ruBnJE5Hy36CpgTShjKObnhOFyimsb0EtE6rj/L1cRvkbFXwB1ROQOONUAfiIwXVWPhSkmU41ZwlEGVS0EvN2qrwUyw9Gtuoi8BXwLnC8i20XkF6GOAedX/e04v+a9tx5eF4Y4WgDzRWQFTkI4T1XDdltqmDUDvhaR5cAS4ENV/SQMcfwGeMN9T7oBT4chBtzGkFfj1CyEnFvL8y6wDKcdRRxh6l5cnW6kbwKGiMhGYANwHHgkHPEYY12bG2OMMSborIbDGGOMMUFnCYcxxhhjgs4SDmOMMcYEnSUcxhhjjAk6SziMMcYYE3SWcMSgkkaODHDZHiIyqQLbnC4it5R3uUgkIo+LyINlzDPYd+A6EXlSRPpXwbani8gtbgzPFJvWTUTO6tNBRH5blWNjVPX6KrIdEVkgIlU6EJ07ou29Pq/7hnOUX2OqG0s4YtNZI0f6TnQ7APJLVbNUdUzxcneMBnPaYJzRgwFQ1cdU9fMqXP9bwK3FyoYBb/qZ97c4g4SdpbT3uhQlrq+KhWo7Xg2Be8uayRgTHJZwxL6vgLbur7n5IvImsFJEaovIayKy0h1w60o481ef+yv7ZRH5DJjpu1Jx/EVE1ojIh0BTn2mPici/3RqWl915zxWRZT7ztBORpe7z8e56VojIc8V3QETq+cS6QkRudsuP+Mxzi4hMd59PF5Gp7v7+ICJXiMirIrLWO09pyxfb9i/dfVkuIu+5PUj2xhmrY4Jbk3SuT83EtSKS6bN8XxH5wH1+jYh8KyLL3JqneiW9aaq6HjgoIpf4FA/F6UbdN74xOGN2zBeR+d79cmtcvgN+IiIjRGSJG+tL3iTEPUZZIrJaRJ4oY33PijM43Oci0tOtgfhBRAa588SLyAT3WK0QkXt89n+BiLwrIutE5A3383DWdkpS0nETkWwRecItXyki7d3yFBGZ55a/JCJbRaQJMB441z0OE9zV1yseW2mxGGMqQVXtEWMP4Ij7NwGYDfwa6IszqFYbd9rvgdfc5+1xumSu7c431y1/HGe0yUQ/2/gZMA+IxzlxHARucacl+8z3d+AG9/l8oJv7/Gmc3imTgfWc7oSuoZ9tPQu84PO6ke9+us9vwemyGWA6zolZcAbSOgx0wUmwl/rEUNLyjwMPus8b+8zzJ+A3Ptu4xWfadHcdCe6xrOuWTwVGAE2ART7l44DH/OzrqfUCY4E/u897Af8u4f3OBpr4vFZgqPu8A/ABUMN9PQW4w/d9ct/DBcAFpazvWvf5P4HPgBpAV+B7t3wU8Kj7vBbOIGJtcD5Ph3DGIIrD6S33Un/bKbZPC3DGASnxuLnLe9+Pe4Fp7vO/AA+7zwe48TcBMoBVPtsoMTZ72MMeVf+wGo7Y5G/kSIAlqrrFfX4pTjKAqq4DtgLn+VnXHFXN91N+OfCWOqO27gS+9Jl2pYh8JyIrgX44w2QDTAPucn9h34pzeeAwTnfL00TkZ4C/MR76A5O9L1T1QGk77/pAVRWne+k9qrpSVT04A1plBLC8V2cR+crdl+E+++KXOl3hfwLcIM5lqOtxkr5eOJdgvnHfm5FA6zK2/TZwi4jEUb7h1otwBtgDZyyP7sC/3e1ehTOcPMBQt9bpP+5+dcS/E+4+gXM8F6ozMNlKTh/La4A73G18BzQG2rnTlqjqdvf4f0/5jn9Zx83bhflSn/VeilsTpE4376V9XioTmzGmHOy6fGzKV2fo9lPcmuKjvkUBrutoKdPO6hdfRGrj/Iruoao5IvI4Ts0JOCfBP+IkJ0tVdb+7TE+cE+EwnHFr+hVfrb9tFSurXWxagfvX4/Pc+9r7uS9tea/pwGBVXS4id+L8Ki7LO8B9wI84tRJ5blX9PFX9eQDLO8E5xy8buAK4GfhJgIseV9Ui97kAM1T1Yd8ZRKQN8CBwsaoecC8nlXQMTrrJG/gcT1X1yOm2PYJT2/Bpse305czjX0T5vnfKOm7edfuutzyXRSoTmzGmHKyGo/pahPOLHRE5D2iFc2mjPMsPc6/dtwCudMu9J6197rX2U3euqOpxnEHwpgKvuduuBySp6kc4jQi7+dnWZziJCO4yjdyne0Skg1sDcFM5YvcKZPn6wC4RqYF7vFx57jR/FuAMz/5LnOQD4F9AHxFp6+5DHfe4l+Ut4M/AZlXdXsI8pcXyBU4tSVN3u8ki0hpogJNMHhKRZjhDugeyvpJ8CvzaPU6IyHnijF5bmkC2U5Hj9jVOexdE5BrA+3mpyH4ZY6qIJRzV1xQg3r1U8A5wp6oWlLGMr38CG3Gq1acCCwFU9SDwN7f8fZzRXH29gVOz8Jn7uj4wV5xRRhcCv/OzrT8BjcRphLqc08nNQ8BcnBqTXeWI3SuQ5f8H5xLBPGCdT/nbwFhxGtye67uAW7swF+ckPtctywXuBN5y9/VfOG1nyjIL53LH26XM8zLwsb/Gl6q6BngU+Mzd7jyghaoux7mUshp4FfgmkPWVYhrOkPTLRGQV8BJl1xaUuZ0KHrcngGvcy0XX4ry3eW6N2jfu52hCqWswxlQ5Gy3WhJQ4/Vskqer/hDsWE5tEpBZQpKqFIvITYGrxS4zGmNCz65UmZETkn8C5nN1Gw5iq1ArIdC+VncC5tGWMCTOr4TDGGGNM0FkbDmOMMcYEnSUcxhhjjAk6SziMMcYYE3SWcBhjjDEm6CzhMMYYY0zQ/X9lyJpmaABZXwAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_clustered_stacked([df1, df2, df3],[\"LABELS\", \"PREDICTIONS\", \"STRICT PREDICTIONS\"], title=\"Labels and predictions by IV treatment duration\",  H=\"/\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'MIMIC labels and predictions by IV treatment duration'}, xlabel='Proir days cumulative IV treatment length', ylabel='Count'>"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_clustered_stacked([df1, df2],[\"LABELS\", \"BASELINE PREDICTIONS\"], title=\"MIMIC labels and predictions by IV treatment duration\",  H=\"\\\\\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# eicu"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# get list \n",
    "eicu_data_list = eicu_data.stay_id.unique().tolist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/tmp/ipykernel_17702/3531519620.py:11: DtypeWarning: Columns (6) have mixed types. Specify dtype option on import or set low_memory=False.\n",
      "  lab = pd.concat([chunk[chunk['patientunitstayid'].isin(eicu_data_list)] for chunk in iter_csv])\n"
     ]
    }
   ],
   "source": [
    "iter_csv = pd.read_csv(r\"eicu-collaborative-research-database-2.0/vitalPeriodic.csv\", iterator=True, chunksize=10000000)\n",
    "vitalperiodic = pd.concat([chunk[chunk['patientunitstayid'].isin(eicu_data_list)] for chunk in iter_csv])\n",
    "\n",
    "iter_csv = pd.read_csv(r\"eicu-collaborative-research-database-2.0/vitalAperiodic.csv\", iterator=True, chunksize=10000000)\n",
    "vitalaperiodic = pd.concat([chunk[chunk['patientunitstayid'].isin(eicu_data_list)] for chunk in iter_csv])\n",
    "\n",
    "iter_csv = pd.read_csv(r\"eicu-collaborative-research-database-2.0/physicalExam.csv\", iterator=True, chunksize=10000000)\n",
    "physicalexam = pd.concat([chunk[chunk['patientunitstayid'].isin(eicu_data_list)] for chunk in iter_csv])\n",
    "\n",
    "iter_csv = pd.read_csv(r\"eicu-collaborative-research-database-2.0/lab.csv\", iterator=True, chunksize=10000000)\n",
    "lab = pd.concat([chunk[chunk['patientunitstayid'].isin(eicu_data_list)] for chunk in iter_csv])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get unitadmittime24\n",
    "# Note unit admit date = 02 Sep 2022\n",
    "iter_csv = pd.read_csv(r\"eicu-collaborative-research-database-2.0/patient.csv\", iterator=True, chunksize=10000000)\n",
    "patients = pd.concat([chunk[chunk['patientunitstayid'].isin(eicu_data_list)] for chunk in iter_csv])\n",
    "\n",
    "# Select relevant columns\n",
    "patients = patients[['patientunitstayid', 'unitadmittime24']]\n",
    "# Convert unit admit time to day (day 0)\n",
    "patients['unitadmittime24'] = pd.to_datetime(patients['unitadmittime24'])\n",
    "# Set start date\n",
    "patients['unitadmittime24'] = patients['unitadmittime24'].apply(lambda t: t.replace(year=2022, month=9, day=2))\n",
    "\n",
    "# Merge\n",
    "vitalperiodic = pd.merge(patients, vitalperiodic, on=['patientunitstayid'])\n",
    "vitalaperiodic = pd.merge(patients, vitalaperiodic, on=['patientunitstayid'])\n",
    "physicalexam = pd.merge(patients, physicalexam, on=['patientunitstayid'])\n",
    "lab = pd.merge(patients, lab, on=['patientunitstayid'])\n",
    "\n",
    "# Create observationtime\n",
    "vitalperiodic['observationoffset'] = pd.to_timedelta(vitalperiodic['observationoffset'], unit='min') # Convert to timedelta\n",
    "vitalperiodic['observationtime'] = vitalperiodic['unitadmittime24'] + vitalperiodic['observationoffset']\n",
    "\n",
    "vitalaperiodic['observationoffset'] = pd.to_timedelta(vitalaperiodic['observationoffset'], unit='min') # Convert to timedelta\n",
    "vitalaperiodic['observationtime'] = vitalaperiodic['unitadmittime24'] + vitalaperiodic['observationoffset']\n",
    "\n",
    "physicalexam['physicalexamoffset'] = pd.to_timedelta(physicalexam['physicalexamoffset'], unit='min') # Convert to timedelta\n",
    "physicalexam['observationtime'] = physicalexam['unitadmittime24'] + physicalexam['physicalexamoffset']\n",
    "\n",
    "lab['labresultoffset'] = pd.to_timedelta(lab['labresultoffset'], unit='min') # Convert to timedelta\n",
    "lab['observationtime'] = lab['unitadmittime24'] + lab['labresultoffset']"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "physicalexam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Are also GCS 'Level of Consciousness' ones but not including as not included in model\n",
    "# physicalexam[physicalexam.physicalexampath.str.contains('GCS', case=False)]['physicalexampath'].value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "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>physicalexampath</th>\n",
       "      <th>patientunitstayid</th>\n",
       "      <th>date</th>\n",
       "      <th>gcs_eye_opening</th>\n",
       "      <th>gcs_motor_response</th>\n",
       "      <th>gcs_verbal_response</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>141392</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>141470</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>142405</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>143068</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>143157</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2515</th>\n",
       "      <td>3335807</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2516</th>\n",
       "      <td>3340875</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2517</th>\n",
       "      <td>3340875</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2518</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2519</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>2520 rows × 5 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "physicalexampath  patientunitstayid        date  gcs_eye_opening  \\\n",
       "0                            141392  2022-09-02              4.0   \n",
       "1                            141470  2022-09-02              4.0   \n",
       "2                            142405  2022-09-02              4.0   \n",
       "3                            143068  2022-09-02              4.0   \n",
       "4                            143157  2022-09-02              4.0   \n",
       "...                             ...         ...              ...   \n",
       "2515                        3335807  2022-09-02              4.0   \n",
       "2516                        3340875  2022-09-02              4.0   \n",
       "2517                        3340875  2022-09-04              4.0   \n",
       "2518                        3341168  2022-09-03              4.0   \n",
       "2519                        3346897  2022-09-02              4.0   \n",
       "\n",
       "physicalexampath  gcs_motor_response  gcs_verbal_response  \n",
       "0                                6.0                  5.0  \n",
       "1                                6.0                  5.0  \n",
       "2                                6.0                  5.0  \n",
       "3                                6.0                  5.0  \n",
       "4                                6.0                  5.0  \n",
       "...                              ...                  ...  \n",
       "2515                             6.0                  5.0  \n",
       "2516                             6.0                  4.0  \n",
       "2517                             6.0                  5.0  \n",
       "2518                             6.0                  4.0  \n",
       "2519                             6.0                  5.0  \n",
       "\n",
       "[2520 rows x 5 columns]"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Filter for GCS\n",
    "physicalexam = physicalexam[physicalexam.physicalexampath.str.contains('GCS')]\n",
    "score = 'Motor Score|Verbal Score|Eyes Score'\n",
    "# Filter for the three scores\n",
    "physicalexam = physicalexam[physicalexam.physicalexampath.str.contains(score)]\n",
    "# Remame\n",
    "physicalexam.loc[physicalexam['physicalexampath'].str.contains('Eyes Score'), 'physicalexampath'] = 'gcs - eye opening'\n",
    "physicalexam.loc[physicalexam['physicalexampath'].str.contains('Verbal Score'), 'physicalexampath'] = 'gcs - verbal response'\n",
    "physicalexam.loc[physicalexam['physicalexampath'].str.contains('Motor Score'), 'physicalexampath'] = 'gcs - motor response'\n",
    "# Order\n",
    "physicalexam = physicalexam.sort_values(by=['patientunitstayid', 'observationtime'])\n",
    "# Create date column \n",
    "physicalexam['date'] =  pd.to_datetime(physicalexam['observationtime']).dt.date\n",
    "# Create hour column\n",
    "#physicalexam['hour'] =  pd.to_datetime(physicalexam['observationtime']).dt.hour # Remove hour from group by as dont need here\n",
    "# Pivot\n",
    "physicalexam_pivoted = pd.pivot_table(physicalexam, index=['patientunitstayid', 'date'], columns=['physicalexampath'], values=['physicalexamvalue'])\n",
    "physicalexam_pivoted.columns = physicalexam_pivoted.columns.droplevel()\n",
    "physicalexam_pivoted = physicalexam_pivoted.reset_index()\n",
    "physicalexam_pivoted.columns = physicalexam_pivoted.columns.str.replace(' ','_') \n",
    "physicalexam_pivoted.rename(columns={'gcs_-_eye_opening': 'gcs_eye_opening', 'gcs_-_motor_response':'gcs_motor_response',\t'gcs_-_verbal_response':'gcs_verbal_response'}, inplace=True)\n",
    "physicalexam_pivoted"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "vitalperiodic / vitalaperiodic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "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>patientunitstayid</th>\n",
       "      <th>date</th>\n",
       "      <th>temperature</th>\n",
       "      <th>sao2</th>\n",
       "      <th>heartrate</th>\n",
       "      <th>respiration</th>\n",
       "      <th>etco2</th>\n",
       "      <th>blood_pressure_systolic</th>\n",
       "      <th>blood_pressure_diastolic</th>\n",
       "      <th>blood_pressure_mean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.437500</td>\n",
       "      <td>91.391304</td>\n",
       "      <td>22.739130</td>\n",
       "      <td>NaN</td>\n",
       "      <td>114.000000</td>\n",
       "      <td>65.000000</td>\n",
       "      <td>83.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.449799</td>\n",
       "      <td>81.465863</td>\n",
       "      <td>24.242915</td>\n",
       "      <td>NaN</td>\n",
       "      <td>134.166667</td>\n",
       "      <td>72.500000</td>\n",
       "      <td>96.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>141392</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>99.458564</td>\n",
       "      <td>74.320442</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>122.370370</td>\n",
       "      <td>59.592593</td>\n",
       "      <td>82.259259</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>141470</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.632479</td>\n",
       "      <td>105.132075</td>\n",
       "      <td>28.031447</td>\n",
       "      <td>NaN</td>\n",
       "      <td>101.611111</td>\n",
       "      <td>55.888889</td>\n",
       "      <td>75.055556</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>142405</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>94.485294</td>\n",
       "      <td>82.925000</td>\n",
       "      <td>26.421053</td>\n",
       "      <td>NaN</td>\n",
       "      <td>121.787879</td>\n",
       "      <td>72.181818</td>\n",
       "      <td>91.030303</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6606</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.220280</td>\n",
       "      <td>98.846154</td>\n",
       "      <td>34.909091</td>\n",
       "      <td>NaN</td>\n",
       "      <td>114.275000</td>\n",
       "      <td>70.050000</td>\n",
       "      <td>86.250000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6607</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.780220</td>\n",
       "      <td>88.835165</td>\n",
       "      <td>29.890110</td>\n",
       "      <td>NaN</td>\n",
       "      <td>121.457143</td>\n",
       "      <td>69.600000</td>\n",
       "      <td>89.857143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6608</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.083333</td>\n",
       "      <td>84.722222</td>\n",
       "      <td>26.333333</td>\n",
       "      <td>NaN</td>\n",
       "      <td>101.100000</td>\n",
       "      <td>55.300000</td>\n",
       "      <td>73.400000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6609</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.232639</td>\n",
       "      <td>80.093750</td>\n",
       "      <td>21.498099</td>\n",
       "      <td>0.0</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>54.320000</td>\n",
       "      <td>71.853333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6610</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.594937</td>\n",
       "      <td>78.396624</td>\n",
       "      <td>18.662447</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.562500</td>\n",
       "      <td>54.062500</td>\n",
       "      <td>70.520833</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>6611 rows × 10 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      patientunitstayid        date  temperature       sao2   heartrate  \\\n",
       "0                141196  2022-09-02          NaN  96.437500   91.391304   \n",
       "1                141196  2022-09-03          NaN  95.449799   81.465863   \n",
       "2                141392  2022-09-02          NaN  99.458564   74.320442   \n",
       "3                141470  2022-09-02          NaN  96.632479  105.132075   \n",
       "4                142405  2022-09-02          NaN  94.485294   82.925000   \n",
       "...                 ...         ...          ...        ...         ...   \n",
       "6606            3341168  2022-09-03          NaN  95.220280   98.846154   \n",
       "6607            3341168  2022-09-04          NaN  95.780220   88.835165   \n",
       "6608            3346897  2022-09-02          NaN  96.083333   84.722222   \n",
       "6609            3346897  2022-09-03          NaN  96.232639   80.093750   \n",
       "6610            3346897  2022-09-04          NaN  96.594937   78.396624   \n",
       "\n",
       "      respiration  etco2  blood_pressure_systolic  blood_pressure_diastolic  \\\n",
       "0       22.739130    NaN               114.000000                 65.000000   \n",
       "1       24.242915    NaN               134.166667                 72.500000   \n",
       "2             NaN    NaN               122.370370                 59.592593   \n",
       "3       28.031447    NaN               101.611111                 55.888889   \n",
       "4       26.421053    NaN               121.787879                 72.181818   \n",
       "...           ...    ...                      ...                       ...   \n",
       "6606    34.909091    NaN               114.275000                 70.050000   \n",
       "6607    29.890110    NaN               121.457143                 69.600000   \n",
       "6608    26.333333    NaN               101.100000                 55.300000   \n",
       "6609    21.498099    0.0               100.000000                 54.320000   \n",
       "6610    18.662447    NaN                97.562500                 54.062500   \n",
       "\n",
       "      blood_pressure_mean  \n",
       "0               83.000000  \n",
       "1               96.500000  \n",
       "2               82.259259  \n",
       "3               75.055556  \n",
       "4               91.030303  \n",
       "...                   ...  \n",
       "6606            86.250000  \n",
       "6607            89.857143  \n",
       "6608            73.400000  \n",
       "6609            71.853333  \n",
       "6610            70.520833  \n",
       "\n",
       "[6611 rows x 10 columns]"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Select columns\n",
    "vitalperiodic = vitalperiodic[['patientunitstayid', 'observationtime', 'temperature', 'sao2', 'heartrate', 'respiration', 'etco2', 'systemicsystolic', 'systemicdiastolic', 'systemicmean']]\n",
    "vitalaperiodic = vitalaperiodic[['patientunitstayid', 'observationtime', 'noninvasivesystolic', 'noninvasivediastolic', 'noninvasivemean']]\n",
    "\n",
    "# Merge\n",
    "vital = pd.merge(vitalperiodic, vitalaperiodic, on=['patientunitstayid', 'observationtime'], how='outer')\n",
    "\n",
    "# Create combined bp columns\n",
    "vital['blood pressure systolic'] = vital[['systemicsystolic', 'noninvasivesystolic']].mean(axis=1)\n",
    "vital['blood pressure diastolic'] = vital[['systemicdiastolic', 'noninvasivediastolic']].mean(axis=1)\n",
    "vital['blood pressure mean'] = vital[['systemicmean', 'noninvasivemean']].mean(axis=1)\n",
    "\n",
    "# Order\n",
    "vital = vital.sort_values(by=['patientunitstayid', 'observationtime'])\n",
    "# Create date column \n",
    "vital['date'] =  pd.to_datetime(vital['observationtime']).dt.date\n",
    "# Create hour column\n",
    "vital['hour'] =  pd.to_datetime(vital['observationtime']).dt.hour\n",
    "# Select columns\n",
    "final_vital = vital[['patientunitstayid', 'date', 'temperature', 'sao2', 'heartrate', 'respiration', 'etco2', 'blood pressure systolic', 'blood pressure diastolic', 'blood pressure mean']]\n",
    "final_vital = final_vital.groupby(['patientunitstayid', 'date']).mean()\n",
    "final_vital.reset_index(inplace=True)\n",
    "final_vital.columns = final_vital.columns.str.replace(' ','_') \n",
    "final_vital"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "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>patientunitstayid</th>\n",
       "      <th>observationtime</th>\n",
       "      <th>temperature</th>\n",
       "      <th>sao2</th>\n",
       "      <th>heartrate</th>\n",
       "      <th>respiration</th>\n",
       "      <th>systemicsystolic</th>\n",
       "      <th>systemicdiastolic</th>\n",
       "      <th>systemicmean</th>\n",
       "      <th>noninvasivesystolic</th>\n",
       "      <th>noninvasivediastolic</th>\n",
       "      <th>noninvasivemean</th>\n",
       "      <th>blood pressure systolic</th>\n",
       "      <th>blood pressure diastolic</th>\n",
       "      <th>blood pressure mean</th>\n",
       "      <th>date</th>\n",
       "      <th>hour</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>257</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-02 22:05:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>265</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-02 22:10:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>267</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-02 22:15:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.0</td>\n",
       "      <td>90.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>259</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-02 22:20:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>98.0</td>\n",
       "      <td>89.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>22</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>253</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-02 22:25:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>98.0</td>\n",
       "      <td>89.0</td>\n",
       "      <td>24.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>22</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1473674</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-04 19:20:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.0</td>\n",
       "      <td>81.0</td>\n",
       "      <td>37.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1473686</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-04 19:25:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>98.0</td>\n",
       "      <td>80.0</td>\n",
       "      <td>25.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1473690</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-04 19:30:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.0</td>\n",
       "      <td>84.0</td>\n",
       "      <td>28.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1473698</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-04 19:35:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>97.0</td>\n",
       "      <td>79.0</td>\n",
       "      <td>18.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1473678</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-04 19:40:00</td>\n",
       "      <td>NaN</td>\n",
       "      <td>100.0</td>\n",
       "      <td>77.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>19</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1636351 rows × 17 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "         patientunitstayid     observationtime  temperature   sao2  heartrate  \\\n",
       "257                 141196 2022-09-02 22:05:00          NaN   97.0       90.0   \n",
       "265                 141196 2022-09-02 22:10:00          NaN   97.0       90.0   \n",
       "267                 141196 2022-09-02 22:15:00          NaN   97.0       90.0   \n",
       "259                 141196 2022-09-02 22:20:00          NaN   98.0       89.0   \n",
       "253                 141196 2022-09-02 22:25:00          NaN   98.0       89.0   \n",
       "...                    ...                 ...          ...    ...        ...   \n",
       "1473674            3346897 2022-09-04 19:20:00          NaN   97.0       81.0   \n",
       "1473686            3346897 2022-09-04 19:25:00          NaN   98.0       80.0   \n",
       "1473690            3346897 2022-09-04 19:30:00          NaN   97.0       84.0   \n",
       "1473698            3346897 2022-09-04 19:35:00          NaN   97.0       79.0   \n",
       "1473678            3346897 2022-09-04 19:40:00          NaN  100.0       77.0   \n",
       "\n",
       "         respiration  systemicsystolic  systemicdiastolic  systemicmean  \\\n",
       "257             24.0               NaN                NaN           NaN   \n",
       "265             24.0               NaN                NaN           NaN   \n",
       "267             24.0               NaN                NaN           NaN   \n",
       "259             23.0               NaN                NaN           NaN   \n",
       "253             24.0               NaN                NaN           NaN   \n",
       "...              ...               ...                ...           ...   \n",
       "1473674         37.0               NaN                NaN           NaN   \n",
       "1473686         25.0               NaN                NaN           NaN   \n",
       "1473690         28.0               NaN                NaN           NaN   \n",
       "1473698         18.0               NaN                NaN           NaN   \n",
       "1473678         16.0               NaN                NaN           NaN   \n",
       "\n",
       "         noninvasivesystolic  noninvasivediastolic  noninvasivemean  \\\n",
       "257                      NaN                   NaN              NaN   \n",
       "265                      NaN                   NaN              NaN   \n",
       "267                      NaN                   NaN              NaN   \n",
       "259                      NaN                   NaN              NaN   \n",
       "253                      NaN                   NaN              NaN   \n",
       "...                      ...                   ...              ...   \n",
       "1473674                  NaN                   NaN              NaN   \n",
       "1473686                  NaN                   NaN              NaN   \n",
       "1473690                  NaN                   NaN              NaN   \n",
       "1473698                  NaN                   NaN              NaN   \n",
       "1473678                  NaN                   NaN              NaN   \n",
       "\n",
       "         blood pressure systolic  blood pressure diastolic  \\\n",
       "257                          NaN                       NaN   \n",
       "265                          NaN                       NaN   \n",
       "267                          NaN                       NaN   \n",
       "259                          NaN                       NaN   \n",
       "253                          NaN                       NaN   \n",
       "...                          ...                       ...   \n",
       "1473674                      NaN                       NaN   \n",
       "1473686                      NaN                       NaN   \n",
       "1473690                      NaN                       NaN   \n",
       "1473698                      NaN                       NaN   \n",
       "1473678                      NaN                       NaN   \n",
       "\n",
       "         blood pressure mean        date  hour  \n",
       "257                      NaN  2022-09-02    22  \n",
       "265                      NaN  2022-09-02    22  \n",
       "267                      NaN  2022-09-02    22  \n",
       "259                      NaN  2022-09-02    22  \n",
       "253                      NaN  2022-09-02    22  \n",
       "...                      ...         ...   ...  \n",
       "1473674                  NaN  2022-09-04    19  \n",
       "1473686                  NaN  2022-09-04    19  \n",
       "1473690                  NaN  2022-09-04    19  \n",
       "1473698                  NaN  2022-09-04    19  \n",
       "1473678                  NaN  2022-09-04    19  \n",
       "\n",
       "[1636351 rows x 17 columns]"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "vital"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "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>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>iv_flag</th>\n",
       "      <th>first_po_flag</th>\n",
       "      <th>po_flag</th>\n",
       "      <th>iv_treatment_length</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>141392</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>141470</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>142405</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>143068</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4942</th>\n",
       "      <td>3335807</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4943</th>\n",
       "      <td>3340875</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4944</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4945</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4946</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4947 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      stay_id        date  iv_flag  first_po_flag  po_flag  \\\n",
       "0      141196  2022-09-03      1.0            NaN        0   \n",
       "1      141392  2022-09-02      1.0            NaN        0   \n",
       "2      141470  2022-09-02      1.0            NaN        0   \n",
       "3      142405  2022-09-02      1.0            NaN        0   \n",
       "4      143068  2022-09-02      1.0            NaN        0   \n",
       "...       ...         ...      ...            ...      ...   \n",
       "4942  3335807  2022-09-02      1.0            NaN        0   \n",
       "4943  3340875  2022-09-04      1.0            NaN        0   \n",
       "4944  3341168  2022-09-03      1.0            NaN        0   \n",
       "4945  3341168  2022-09-04      1.0            NaN        0   \n",
       "4946  3346897  2022-09-02      1.0            NaN        0   \n",
       "\n",
       "      iv_treatment_length  \n",
       "0                       0  \n",
       "1                       0  \n",
       "2                       0  \n",
       "3                       0  \n",
       "4                       0  \n",
       "...                   ...  \n",
       "4942                    0  \n",
       "4943                    0  \n",
       "4944                    0  \n",
       "4945                    1  \n",
       "4946                    0  \n",
       "\n",
       "[4947 rows x 6 columns]"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eicu_data"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "lab"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "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>patientunitstayid</th>\n",
       "      <th>date</th>\n",
       "      <th>inspired_o2_fraction</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>142405</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>26.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>142405</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>25.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>143068</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>46.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>143157</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>44.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>144427</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1895</th>\n",
       "      <td>3340875</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>86.666667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1896</th>\n",
       "      <td>3340875</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>50.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1897</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>44.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1898</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-06</td>\n",
       "      <td>36.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1899</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>36.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>1900 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      patientunitstayid        date  inspired_o2_fraction\n",
       "0                142405  2022-09-02             26.500000\n",
       "1                142405  2022-09-03             25.000000\n",
       "2                143068  2022-09-02             46.666667\n",
       "3                143157  2022-09-02             44.000000\n",
       "4                144427  2022-09-02                   NaN\n",
       "...                 ...         ...                   ...\n",
       "1895            3340875  2022-09-02             86.666667\n",
       "1896            3340875  2022-09-03             50.000000\n",
       "1897            3341168  2022-09-03             44.000000\n",
       "1898            3341168  2022-09-06             36.000000\n",
       "1899            3346897  2022-09-02             36.000000\n",
       "\n",
       "[1900 rows x 3 columns]"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Filter for FiO2\n",
    "filtered_lab = lab[lab['labname'].str.contains('FiO2', case=False)]\n",
    "# Select columns\n",
    "filtered_lab = filtered_lab[['patientunitstayid', 'observationtime', 'labresult']]\n",
    "# Order\n",
    "filtered_lab = filtered_lab.sort_values(by=['patientunitstayid', 'observationtime'])\n",
    "# Create date column \n",
    "filtered_lab['date'] =  pd.to_datetime(filtered_lab['observationtime']).dt.date\n",
    "# Create hour column\n",
    "filtered_lab['hour'] =  pd.to_datetime(filtered_lab['observationtime']).dt.hour\n",
    "# Select columns\n",
    "final_filtered_lab = filtered_lab[['patientunitstayid', 'date', 'labresult']]\n",
    "final_filtered_lab = final_filtered_lab.groupby(['patientunitstayid', 'date']).mean()\n",
    "final_filtered_lab.reset_index(inplace=True)\n",
    "final_filtered_lab.rename(columns={'labresult': 'inspired_o2_fraction'}, inplace=True)\n",
    "final_filtered_lab"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Min Max Temp"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Filter\n",
    "eicu_data2 = vital[['patientunitstayid', 'date', 'temperature']]\n",
    "eicu_data2.rename(columns={'patientunitstayid':'stay_id', 'date':'date2'}, inplace=True)\n",
    "# Merge\n",
    "eicu_data2 = pd.merge(eicu_data, eicu_data2[['stay_id', 'date2', 'temperature']])\n",
    "# Datetime\n",
    "import datetime\n",
    "eicu_data2['date2'] =  pd.to_datetime(eicu_data2['date2']).dt.date + datetime.timedelta(days=1)\n",
    "# Convert to str\n",
    "eicu_data2['date2'] = eicu_data2['date2'].astype(str)\n",
    "eicu_data2['date'] = eicu_data2['date'].astype(str)\n",
    "# Filter\n",
    "eicu_data3 = eicu_data2.loc[eicu_data2['date'] == eicu_data2['date2']]\n",
    "def summary_function(df):\n",
    "    return df['temperature'].agg(['min','max'])\n",
    "# Get max and min\n",
    "eicu_data4 = eicu_data3.groupby(['stay_id', 'date'], as_index=False).apply(summary_function)\n",
    "# Merge\n",
    "eicu_data5 = pd.merge(eicu_data3[['stay_id', 'date', 'iv_flag', 'first_po_flag', 'po_flag', 'iv_treatment_length']], eicu_data4)\n",
    "eicu_data5.drop_duplicates(inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 3716 entries, 0 to 957828\n",
      "Data columns (total 8 columns):\n",
      " #   Column               Non-Null Count  Dtype  \n",
      "---  ------               --------------  -----  \n",
      " 0   stay_id              3716 non-null   int64  \n",
      " 1   date                 3716 non-null   object \n",
      " 2   iv_flag              2914 non-null   float64\n",
      " 3   first_po_flag        1673 non-null   float64\n",
      " 4   po_flag              3716 non-null   int64  \n",
      " 5   iv_treatment_length  3716 non-null   int64  \n",
      " 6   min                  289 non-null    float64\n",
      " 7   max                  289 non-null    float64\n",
      "dtypes: float64(4), int64(3), object(1)\n",
      "memory usage: 261.3+ KB\n"
     ]
    }
   ],
   "source": [
    "eicu_data5.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "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>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>iv_flag</th>\n",
       "      <th>first_po_flag</th>\n",
       "      <th>po_flag</th>\n",
       "      <th>iv_treatment_length</th>\n",
       "      <th>min</th>\n",
       "      <th>max</th>\n",
       "      <th>predictions</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>143068</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>232</th>\n",
       "      <td>143068</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>999</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>523</th>\n",
       "      <td>143157</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>666</th>\n",
       "      <td>143157</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>956609</th>\n",
       "      <td>3247070</td>\n",
       "      <td>2022-09-05</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>956909</th>\n",
       "      <td>3247070</td>\n",
       "      <td>2022-09-06</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>999</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>957197</th>\n",
       "      <td>3247135</td>\n",
       "      <td>2022-09-05</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>1</td>\n",
       "      <td>999</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>957492</th>\n",
       "      <td>3340875</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>957828</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>3716 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "        stay_id        date  iv_flag  first_po_flag  po_flag  \\\n",
       "0        141196  2022-09-03      1.0            NaN        0   \n",
       "24       143068  2022-09-03      NaN            1.0        1   \n",
       "232      143068  2022-09-04      NaN            1.0        1   \n",
       "523      143157  2022-09-03      1.0            NaN        0   \n",
       "666      143157  2022-09-04      1.0            NaN        0   \n",
       "...         ...         ...      ...            ...      ...   \n",
       "956609  3247070  2022-09-05      NaN            1.0        1   \n",
       "956909  3247070  2022-09-06      NaN            1.0        1   \n",
       "957197  3247135  2022-09-05      NaN            1.0        1   \n",
       "957492  3340875  2022-09-04      1.0            NaN        0   \n",
       "957828  3341168  2022-09-04      1.0            NaN        0   \n",
       "\n",
       "        iv_treatment_length  min  max  predictions  \n",
       "0                         0  NaN  NaN            1  \n",
       "24                        1  NaN  NaN            1  \n",
       "232                     999  NaN  NaN            1  \n",
       "523                       1  NaN  NaN            1  \n",
       "666                       2  NaN  NaN            1  \n",
       "...                     ...  ...  ...          ...  \n",
       "956609                    3  NaN  NaN            1  \n",
       "956909                  999  NaN  NaN            1  \n",
       "957197                  999  NaN  NaN            1  \n",
       "957492                    0  NaN  NaN            1  \n",
       "957828                    1  NaN  NaN            1  \n",
       "\n",
       "[3716 rows x 9 columns]"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eicu_data5"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "NEWS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convert to str or does not work\n",
    "eicu_data.date = eicu_data.date.astype(str)\n",
    "final_vital.date = final_vital.date.astype(str)\n",
    "final_filtered_lab.date = final_filtered_lab.date.astype(str)\n",
    "physicalexam_pivoted.date = physicalexam_pivoted.date.astype(str)\n",
    "# Rename\n",
    "final_vital.rename(columns={'patientunitstayid': 'stay_id'}, inplace=True)\n",
    "final_filtered_lab.rename(columns={'patientunitstayid': 'stay_id'}, inplace=True)\n",
    "physicalexam_pivoted.rename(columns={'patientunitstayid': 'stay_id'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Merge pivots\n",
    "eicu_pivoted = pd.merge(eicu_data, final_vital, how='left')\n",
    "eicu_pivoted = pd.merge(eicu_pivoted, final_filtered_lab, how='left')\n",
    "eicu_pivoted = pd.merge(eicu_pivoted, physicalexam_pivoted, how='left')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "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>stay_id</th>\n",
       "      <th>date</th>\n",
       "      <th>iv_flag</th>\n",
       "      <th>first_po_flag</th>\n",
       "      <th>po_flag</th>\n",
       "      <th>iv_treatment_length</th>\n",
       "      <th>temperature</th>\n",
       "      <th>sao2</th>\n",
       "      <th>heartrate</th>\n",
       "      <th>respiration</th>\n",
       "      <th>etco2</th>\n",
       "      <th>blood_pressure_systolic</th>\n",
       "      <th>blood_pressure_diastolic</th>\n",
       "      <th>blood_pressure_mean</th>\n",
       "      <th>inspired_o2_fraction</th>\n",
       "      <th>gcs_eye_opening</th>\n",
       "      <th>gcs_motor_response</th>\n",
       "      <th>gcs_verbal_response</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>141196</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.449799</td>\n",
       "      <td>81.465863</td>\n",
       "      <td>24.242915</td>\n",
       "      <td>NaN</td>\n",
       "      <td>134.166667</td>\n",
       "      <td>72.500000</td>\n",
       "      <td>96.500000</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>141392</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>99.458564</td>\n",
       "      <td>74.320442</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>122.370370</td>\n",
       "      <td>59.592593</td>\n",
       "      <td>82.259259</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>141470</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.632479</td>\n",
       "      <td>105.132075</td>\n",
       "      <td>28.031447</td>\n",
       "      <td>NaN</td>\n",
       "      <td>101.611111</td>\n",
       "      <td>55.888889</td>\n",
       "      <td>75.055556</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>142405</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>94.485294</td>\n",
       "      <td>82.925000</td>\n",
       "      <td>26.421053</td>\n",
       "      <td>NaN</td>\n",
       "      <td>121.787879</td>\n",
       "      <td>72.181818</td>\n",
       "      <td>91.030303</td>\n",
       "      <td>26.500000</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>143068</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>99.456731</td>\n",
       "      <td>129.774038</td>\n",
       "      <td>20.157895</td>\n",
       "      <td>NaN</td>\n",
       "      <td>93.875000</td>\n",
       "      <td>48.004808</td>\n",
       "      <td>62.105769</td>\n",
       "      <td>46.666667</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4942</th>\n",
       "      <td>3335807</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>98.256410</td>\n",
       "      <td>108.974359</td>\n",
       "      <td>24.102564</td>\n",
       "      <td>NaN</td>\n",
       "      <td>144.000000</td>\n",
       "      <td>74.000000</td>\n",
       "      <td>99.666667</td>\n",
       "      <td>28.000000</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4943</th>\n",
       "      <td>3340875</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.815315</td>\n",
       "      <td>112.639640</td>\n",
       "      <td>23.081081</td>\n",
       "      <td>NaN</td>\n",
       "      <td>109.185185</td>\n",
       "      <td>64.777778</td>\n",
       "      <td>81.518519</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4944</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-03</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.220280</td>\n",
       "      <td>98.846154</td>\n",
       "      <td>34.909091</td>\n",
       "      <td>NaN</td>\n",
       "      <td>114.275000</td>\n",
       "      <td>70.050000</td>\n",
       "      <td>86.250000</td>\n",
       "      <td>44.000000</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4945</th>\n",
       "      <td>3341168</td>\n",
       "      <td>2022-09-04</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>95.780220</td>\n",
       "      <td>88.835165</td>\n",
       "      <td>29.890110</td>\n",
       "      <td>NaN</td>\n",
       "      <td>121.457143</td>\n",
       "      <td>69.600000</td>\n",
       "      <td>89.857143</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4946</th>\n",
       "      <td>3346897</td>\n",
       "      <td>2022-09-02</td>\n",
       "      <td>1.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>96.083333</td>\n",
       "      <td>84.722222</td>\n",
       "      <td>26.333333</td>\n",
       "      <td>NaN</td>\n",
       "      <td>101.100000</td>\n",
       "      <td>55.300000</td>\n",
       "      <td>73.400000</td>\n",
       "      <td>36.000000</td>\n",
       "      <td>4.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>4947 rows × 18 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "      stay_id        date  iv_flag  first_po_flag  po_flag  \\\n",
       "0      141196  2022-09-03      1.0            NaN        0   \n",
       "1      141392  2022-09-02      1.0            NaN        0   \n",
       "2      141470  2022-09-02      1.0            NaN        0   \n",
       "3      142405  2022-09-02      1.0            NaN        0   \n",
       "4      143068  2022-09-02      1.0            NaN        0   \n",
       "...       ...         ...      ...            ...      ...   \n",
       "4942  3335807  2022-09-02      1.0            NaN        0   \n",
       "4943  3340875  2022-09-04      1.0            NaN        0   \n",
       "4944  3341168  2022-09-03      1.0            NaN        0   \n",
       "4945  3341168  2022-09-04      1.0            NaN        0   \n",
       "4946  3346897  2022-09-02      1.0            NaN        0   \n",
       "\n",
       "      iv_treatment_length  temperature       sao2   heartrate  respiration  \\\n",
       "0                       0          NaN  95.449799   81.465863    24.242915   \n",
       "1                       0          NaN  99.458564   74.320442          NaN   \n",
       "2                       0          NaN  96.632479  105.132075    28.031447   \n",
       "3                       0          NaN  94.485294   82.925000    26.421053   \n",
       "4                       0          NaN  99.456731  129.774038    20.157895   \n",
       "...                   ...          ...        ...         ...          ...   \n",
       "4942                    0          NaN  98.256410  108.974359    24.102564   \n",
       "4943                    0          NaN  96.815315  112.639640    23.081081   \n",
       "4944                    0          NaN  95.220280   98.846154    34.909091   \n",
       "4945                    1          NaN  95.780220   88.835165    29.890110   \n",
       "4946                    0          NaN  96.083333   84.722222    26.333333   \n",
       "\n",
       "      etco2  blood_pressure_systolic  blood_pressure_diastolic  \\\n",
       "0       NaN               134.166667                 72.500000   \n",
       "1       NaN               122.370370                 59.592593   \n",
       "2       NaN               101.611111                 55.888889   \n",
       "3       NaN               121.787879                 72.181818   \n",
       "4       NaN                93.875000                 48.004808   \n",
       "...     ...                      ...                       ...   \n",
       "4942    NaN               144.000000                 74.000000   \n",
       "4943    NaN               109.185185                 64.777778   \n",
       "4944    NaN               114.275000                 70.050000   \n",
       "4945    NaN               121.457143                 69.600000   \n",
       "4946    NaN               101.100000                 55.300000   \n",
       "\n",
       "      blood_pressure_mean  inspired_o2_fraction  gcs_eye_opening  \\\n",
       "0               96.500000                   NaN              NaN   \n",
       "1               82.259259                   NaN              4.0   \n",
       "2               75.055556                   NaN              4.0   \n",
       "3               91.030303             26.500000              4.0   \n",
       "4               62.105769             46.666667              4.0   \n",
       "...                   ...                   ...              ...   \n",
       "4942            99.666667             28.000000              4.0   \n",
       "4943            81.518519                   NaN              4.0   \n",
       "4944            86.250000             44.000000              4.0   \n",
       "4945            89.857143                   NaN              NaN   \n",
       "4946            73.400000             36.000000              4.0   \n",
       "\n",
       "      gcs_motor_response  gcs_verbal_response  \n",
       "0                    NaN                  NaN  \n",
       "1                    6.0                  5.0  \n",
       "2                    6.0                  5.0  \n",
       "3                    6.0                  5.0  \n",
       "4                    6.0                  5.0  \n",
       "...                  ...                  ...  \n",
       "4942                 6.0                  5.0  \n",
       "4943                 6.0                  5.0  \n",
       "4944                 6.0                  4.0  \n",
       "4945                 NaN                  NaN  \n",
       "4946                 6.0                  5.0  \n",
       "\n",
       "[4947 rows x 18 columns]"
      ]
     },
     "execution_count": 65,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eicu_pivoted"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "def NEWS2_function_eicu(respiratory_rate, arterial_co2_pressure, o2_saturation_pulseoxymetry, inspired_o2_fraction, temperature, blood_pressure_systolic, heart_rate, gcs_eye_opening, gcs_motor_response, gcs_verbal_response):\n",
    "    score = 0\n",
    "\n",
    "    if respiratory_rate <= 8:\n",
    "        score += 3\n",
    "    elif respiratory_rate < 12:\n",
    "        score += 1\n",
    "    elif respiratory_rate >= 25:\n",
    "        score += 3\n",
    "    elif respiratory_rate > 20:\n",
    "        score += 2\n",
    "    \n",
    "    if arterial_co2_pressure >= 50:\n",
    "        if o2_saturation_pulseoxymetry <= 83:\n",
    "            score += 3\n",
    "        elif o2_saturation_pulseoxymetry < 86:\n",
    "            score += 2\n",
    "        elif o2_saturation_pulseoxymetry < 88:\n",
    "            score += 1\n",
    "        elif inspired_o2_fraction > 21:\n",
    "            if o2_saturation_pulseoxymetry >= 97:\n",
    "                score += 3\n",
    "            elif o2_saturation_pulseoxymetry > 95:\n",
    "                score += 2\n",
    "            elif o2_saturation_pulseoxymetry > 93:\n",
    "                score += 1\n",
    "    else:\n",
    "        if o2_saturation_pulseoxymetry <= 91:\n",
    "            score += 3\n",
    "        elif o2_saturation_pulseoxymetry < 94:\n",
    "            score += 2\n",
    "        elif o2_saturation_pulseoxymetry < 96:\n",
    "            score += 1\n",
    "\n",
    "    if inspired_o2_fraction > 21:\n",
    "        score += 2\n",
    "    \n",
    "    if temperature <= 35:\n",
    "        score += 3\n",
    "    elif temperature <= 36:\n",
    "        score += 1\n",
    "    elif temperature > 39:\n",
    "        score += 2\n",
    "    elif temperature > 38:\n",
    "        score += 1\n",
    "    \n",
    "    if blood_pressure_systolic <= 90:\n",
    "        score += 3\n",
    "    elif blood_pressure_systolic <= 100:\n",
    "        score += 2\n",
    "    elif blood_pressure_systolic <= 100:\n",
    "        score += 1\n",
    "    elif blood_pressure_systolic >= 220:\n",
    "        score += 3\n",
    "    \n",
    "    if heart_rate <= 40:\n",
    "        score += 3\n",
    "    elif heart_rate <= 50:\n",
    "        score += 1\n",
    "    elif heart_rate >= 131:\n",
    "        score += 3\n",
    "    elif heart_rate >= 111:\n",
    "        score += 2\n",
    "    elif heart_rate >= 91:\n",
    "        score += 1\n",
    "\n",
    "    if gcs_eye_opening < 4:\n",
    "        score += 3\n",
    "    elif gcs_verbal_response < 5:\n",
    "        score += 3\n",
    "    elif gcs_motor_response < 6:\n",
    "        score += 3\n",
    "\n",
    "    return score"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Calculate NEWS 2 score for each\n",
    "eicu_pivoted['news2'] = eicu_pivoted.apply(lambda x: NEWS2_function_eicu(x['respiration'], x['etco2'], x['sao2'], x['inspired_o2_fraction'], x['temperature'], x['blood_pressure_systolic'], x['heartrate'], x['gcs_eye_opening'], x['gcs_motor_response'], x['gcs_verbal_response']), axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "3     900\n",
       "2     879\n",
       "4     732\n",
       "0     644\n",
       "1     631\n",
       "5     499\n",
       "6     286\n",
       "7     197\n",
       "8      97\n",
       "9      40\n",
       "10     25\n",
       "11     13\n",
       "12      2\n",
       "16      1\n",
       "13      1\n",
       "Name: news2, dtype: int64"
      ]
     },
     "execution_count": 70,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eicu_pivoted.news2.value_counts()"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Merge\n",
    "eicu_pivoted2 = pd.merge(eicu_pivoted2, eicu_data5, how='outer')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4947\n"
     ]
    }
   ],
   "source": [
    "# Get baseline predictions \n",
    "cumcount = []\n",
    "count = 1\n",
    "pos = -1\n",
    "flag = 0\n",
    "\n",
    "for x in range(len(eicu_pivoted2)):\n",
    "    pos += 1\n",
    "    if pos == 0:\n",
    "        cumcount.append(np.nan)\n",
    "    elif eicu_pivoted2.iloc[x]['stay_id'] == eicu_pivoted2.iloc[x-1]['stay_id']:\n",
    "        if flag == 1:\n",
    "            cumcount.append(1) # Changed to aways predict switch after initial switch prediction for each stay\n",
    "        else:\n",
    "            if np.isnan(eicu_pivoted2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "                cumcount.append(np.nan)\n",
    "            elif eicu_pivoted2.iloc[x]['news2'] < eicu_pivoted2.iloc[x-1]['news2']:\n",
    "                if eicu_pivoted2.iloc[x]['max'] > 38:\n",
    "                    cumcount.append(0)\n",
    "                elif eicu_pivoted2.iloc[x]['min'] < 36:\n",
    "                    cumcount.append(0)\n",
    "                else:\n",
    "                    cumcount.append(1)\n",
    "                    flag = 1\n",
    "            else:\n",
    "                cumcount.append(0)\n",
    "    elif np.isnan(eicu_pivoted2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "        cumcount.append(np.nan)\n",
    "        flag = 0\n",
    "    elif eicu_pivoted2.iloc[x]['news2'] < eicu_pivoted2.iloc[x-1]['news2']:\n",
    "        if eicu_pivoted2.iloc[x]['max'] > 38:\n",
    "            cumcount.append(0)\n",
    "            flag = 0\n",
    "        elif eicu_pivoted2.iloc[x]['min'] < 36:\n",
    "            cumcount.append(0)\n",
    "            flag = 0\n",
    "        else:\n",
    "            cumcount.append(1)\n",
    "            flag = 1\n",
    "    else:\n",
    "        cumcount.append(0)\n",
    "        flag = 0\n",
    "\n",
    "print(len(cumcount))\n",
    "eicu_pivoted2['predictions'] = cumcount"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4947\n"
     ]
    }
   ],
   "source": [
    "# Get strictbaseline predictions \n",
    "cumcount = []\n",
    "count = 1\n",
    "pos = -1\n",
    "flag = 0\n",
    "\n",
    "for x in range(len(eicu_pivoted2)):\n",
    "    pos += 1\n",
    "    if pos == 0:\n",
    "        cumcount.append(np.nan)\n",
    "    elif eicu_pivoted2.iloc[x]['stay_id'] == eicu_pivoted2.iloc[x-1]['stay_id']:\n",
    "        if flag == 1:\n",
    "            cumcount.append(1) # Changed to aways predict switch after initial switch prediction for each stay\n",
    "        else:\n",
    "            if np.isnan(eicu_pivoted2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "                cumcount.append(np.nan)\n",
    "            elif eicu_pivoted2.iloc[x]['news2'] < eicu_pivoted2.iloc[x-1]['news2']:\n",
    "                if eicu_pivoted2.iloc[x]['news2'] < 7: # Add in threshold of NEWS less than 7\n",
    "                    if eicu_pivoted2.iloc[x]['max'] > 38:\n",
    "                        cumcount.append(0)\n",
    "                    elif eicu_pivoted2.iloc[x]['min'] < 36:\n",
    "                        cumcount.append(0)\n",
    "                    else:\n",
    "                        cumcount.append(1)\n",
    "                        flag = 1\n",
    "                else:\n",
    "                    cumcount.append(0)\n",
    "            else:\n",
    "                cumcount.append(0)\n",
    "    elif np.isnan(eicu_pivoted2.iloc[x]['max']):# == chartevents_pivoted2.iloc[x]['max']: # Check if nan\n",
    "        cumcount.append(np.nan)\n",
    "        flag = 0\n",
    "    elif eicu_pivoted2.iloc[x]['news2'] < eicu_pivoted2.iloc[x-1]['news2']:\n",
    "        if eicu_pivoted2.iloc[x]['news2'] < 7: # Add in threshold of NEWS less than 7\n",
    "            if eicu_pivoted2.iloc[x]['max'] > 38:\n",
    "                cumcount.append(0)\n",
    "                flag = 0\n",
    "            elif eicu_pivoted2.iloc[x]['min'] < 36:\n",
    "                cumcount.append(0)\n",
    "                flag = 0\n",
    "            else:\n",
    "                cumcount.append(1)\n",
    "                flag = 1\n",
    "        else:\n",
    "            cumcount.append(0)\n",
    "            flag = 0\n",
    "    else:\n",
    "        cumcount.append(0)\n",
    "        flag = 0\n",
    "\n",
    "print(len(cumcount))\n",
    "eicu_pivoted2['strict_predictions'] = cumcount"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0    219\n",
       "1.0     91\n",
       "Name: predictions, dtype: int64"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "0.0    219\n",
       "1.0     91\n",
       "Name: strict_predictions, dtype: int64"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "eicu_pivoted2.predictions.value_counts()\n",
    "eicu_pivoted2.strict_predictions.value_counts()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get prediction df\n",
    "prediction_df2 = eicu_pivoted2[['stay_id', 'date', 'iv_flag', 'first_po_flag', 'po_flag', 'iv_treatment_length', 'predictions', 'strict_predictions']]\n",
    "prediction_df2 = prediction_df2[prediction_df2['predictions'].notna()]\n",
    "prediction_df2 = prediction_df2[prediction_df2['strict_predictions'].notna()]"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUROC: 0.5505508173418621\n",
      "Accuracy: 0.6741935483870968\n",
      "Recall: 0.38095238095238093\n",
      "Precision: 0.17582417582417584\n",
      "F1: 0.2406015037593985\n",
      "AUPRC: 0.15085160615114535\n",
      "CM: [[193  75]\n",
      " [ 26  16]]\n",
      "TPR: 0.38095238095238093\n",
      "FPR: 0.2798507462686567\n"
     ]
    }
   ],
   "source": [
    "get_results(prediction_df2.po_flag, prediction_df2.predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "AUROC: 0.5505508173418621\n",
      "Accuracy: 0.6741935483870968\n",
      "Recall: 0.38095238095238093\n",
      "Precision: 0.17582417582417584\n",
      "F1: 0.2406015037593985\n",
      "AUPRC: 0.15085160615114535\n",
      "CM: [[193  75]\n",
      " [ 26  16]]\n",
      "TPR: 0.38095238095238093\n",
      "FPR: 0.2798507462686567\n"
     ]
    }
   ],
   "source": [
    "get_results(prediction_df2.po_flag, prediction_df2.strict_predictions)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "treatment_length_df = pd.DataFrame()\n",
    "\n",
    "for i in prediction_df2.iv_treatment_length.unique():\n",
    "    print(i)\n",
    "    temp_test_data = prediction_df2[prediction_df2['iv_treatment_length'] == i]\n",
    "    test_labels = temp_test_data.po_flag\n",
    "    lower_bound_test_predictions = temp_test_data.predictions\n",
    "    upper_bound_test_predictions = temp_test_data.strict_predictions\n",
    "\n",
    "    print(np.unique(test_labels, return_counts=True))\n",
    "    print(np.unique(lower_bound_test_predictions, return_counts=True))\n",
    "\n",
    "    label_values, label_counts = np.unique(test_labels, return_counts=True)\n",
    "    lower_bound_values, lower_bound_counts =np.unique(lower_bound_test_predictions, return_counts=True)\n",
    "    upper_bound_values, upper_bound_counts =np.unique(upper_bound_test_predictions, return_counts=True)\n",
    "\n",
    "    label_0 = 0\n",
    "    label_1 = 0\n",
    "    for x in range(len(label_values)):\n",
    "        if label_values[x] == 0:\n",
    "            label_0 = label_counts[x]\n",
    "        elif label_values[x] == 1:\n",
    "            label_1 = label_counts[x]\n",
    "    prediction_0 = 0\n",
    "    prediction_1 = 0\n",
    "    for x in range(len(lower_bound_values)):\n",
    "        if lower_bound_values[x] == 0:\n",
    "            prediction_0 = lower_bound_counts[x]\n",
    "        elif lower_bound_values[x] == 1:\n",
    "            prediction_1 = lower_bound_counts[x]\n",
    "    ub_prediction_0 = 0\n",
    "    ub_prediction_1 = 0\n",
    "    for x in range(len(upper_bound_values)):\n",
    "        if upper_bound_values[x] == 0:\n",
    "            ub_prediction_0 = upper_bound_counts[x]\n",
    "        elif upper_bound_values[x] == 1:\n",
    "            ub_prediction_1 = upper_bound_counts[x]\n",
    "    \n",
    "    # Lower bound\n",
    "    try:\n",
    "        test_auroc2 = roc_auc_score(test_labels, lower_bound_test_predictions)\n",
    "    except:\n",
    "        test_auroc2 = np.nan\n",
    "    #print('Test AUROC result 2:', test_auroc2)\n",
    "    test_accuracy2 = accuracy_score(test_labels, lower_bound_test_predictions)\n",
    "    #assert test_accuracy == test_accuracy2\n",
    "    test_balanced_accuracy = balanced_accuracy_score(test_labels, lower_bound_test_predictions)\n",
    "    test_recall = recall_score(test_labels, lower_bound_test_predictions)\n",
    "    test_precision = precision_score(test_labels, lower_bound_test_predictions)\n",
    "    test_f1 = f1_score(test_labels, lower_bound_test_predictions)\n",
    "    test_auprc = average_precision_score(test_labels, lower_bound_test_predictions)\n",
    "    test_cm = confusion_matrix(test_labels, lower_bound_test_predictions)\n",
    "    if test_cm.shape == (2, 2):\n",
    "        tn, fp, fn, tp = test_cm.ravel()\n",
    "        test_true_positive_rate = (tp / (tp + fn))\n",
    "        test_false_positive_rate = (fp / (fp + tn))\n",
    "    else:\n",
    "        test_true_positive_rate = np.nan\n",
    "        test_false_positive_rate = np.nan\n",
    "\n",
    "    # Upper bound\n",
    "    try:\n",
    "        ub_test_auroc2 = roc_auc_score(test_labels, upper_bound_test_predictions)\n",
    "    except:\n",
    "        ub_test_auroc2 = np.nan\n",
    "    #print('Test ub_AUROC result 2:', ub_test_auroc2)\n",
    "    ub_test_accuracy2 = accuracy_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_balanced_accuracy = balanced_accuracy_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_recall = recall_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_precision = precision_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_f1 = f1_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_auprc = average_precision_score(test_labels, upper_bound_test_predictions)\n",
    "    ub_test_cm = confusion_matrix(test_labels, upper_bound_test_predictions)\n",
    "    if ub_test_cm.shape == (2, 2):\n",
    "        tn, fp, fn, tp = ub_test_cm.ravel()\n",
    "        ub_test_true_positive_rate = (tp / (tp + fn))\n",
    "        ub_test_false_positive_rate = (fp / (fp + tn))\n",
    "    else:\n",
    "        ub_test_true_positive_rate = np.nan\n",
    "        ub_test_false_positive_rate = np.nan\n",
    "    \n",
    "    sub_df = pd.DataFrame([[i, label_0, label_1, prediction_0, prediction_1, test_auroc2, test_balanced_accuracy, test_accuracy2, test_recall, test_precision, test_f1, test_auprc, test_cm, test_true_positive_rate, test_false_positive_rate, ub_prediction_0, ub_prediction_1, ub_test_auroc2, ub_test_balanced_accuracy, ub_test_accuracy2, ub_test_recall, ub_test_precision, ub_test_f1, ub_test_auprc, ub_test_cm, ub_test_true_positive_rate, ub_test_false_positive_rate]])\n",
    "    treatment_length_df = pd.concat([treatment_length_df, sub_df], axis=0, ignore_index=True)\n",
    "treatment_length_df.columns = ['iv_treatment_length', 'label_0', 'label_1', 'prediction_0', 'prediction_1', 'auroc', 'balanced_accuracy', 'accuracy', 'recall', 'precision', 'f1', 'auprc', 'cm', 'tpr', 'fpr', 'ub_prediction_0', 'ub_prediction_1','ub_auroc', 'ub_balanced_accuracy', 'ub_accuracy', 'ub_recall', 'ub_precision', 'ub_f1', 'ub_auprc', 'ub_cm', 'ub_tpr', 'ub_fpr']\n",
    "treatment_length_df.sort_values(by=['iv_treatment_length'], inplace=True)\n",
    "# Set to string\n",
    "treatment_length_df['iv_treatment_length']= treatment_length_df['iv_treatment_length'].astype(str)\n",
    "# Change 999\n",
    "treatment_length_df['iv_treatment_length'] = treatment_length_df['iv_treatment_length'].replace(['999'], 'PO')\n",
    "# Rename\n",
    "treatment_length_df.rename(columns={'iv_treatment_length': 'Proir days cumulative IV treatment length', 'label_0': 'Continue with IV', 'label_1': 'Switch to PO'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "df1 = treatment_length_df[['Proir days cumulative IV treatment length', 'Continue with IV', 'Switch to PO']].set_index(['Proir days cumulative IV treatment length'])\n",
    "df2 = treatment_length_df[['Proir days cumulative IV treatment length', 'prediction_0', 'prediction_1']].set_index(['Proir days cumulative IV treatment length'])\n",
    "df3 = treatment_length_df[['Proir days cumulative IV treatment length', 'ub_prediction_0', 'ub_prediction_1']].set_index(['Proir days cumulative IV treatment length'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'Labels and predictions by IV treatment duration'}, xlabel='Proir days cumulative IV treatment length', ylabel='Count'>"
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_clustered_stacked([df1, df2, df3],[\"LABELS\", \"PREDICTIONS\", \"STRICT PREDICTIONS\"], title=\"Labels and predictions by IV treatment duration\",  H=\"/\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:title={'center':'eICU labels and predictions by IV treatment duration'}, xlabel='Proir days cumulative IV treatment length', ylabel='Count'>"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_clustered_stacked([df1, df2],[\"LABELS\", \"BASELINE PREDICTIONS\"], title=\"eICU labels and predictions by IV treatment duration\",  H=\"\\\\\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "stop",
   "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": "3.9.12"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}