mortality_in_septic_patients.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eJI2-t_8sGWV"
   },
   "source": [
    "# MIMIC: Medical Information Mart for Intensive Care \n",
    "\n",
    "Predicting mortality in patients with sepsis (notebook prepared by Giovanni Angelotti)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "uPNxuFkZt8pI"
   },
   "source": [
    "\n",
    "\n",
    "*   [Documentation](https://mimic.physionet.org)\n",
    "*   Single center: Beth Israel Deaconess Medical Center\n",
    "*   Detailed in-ICU information \n",
    "*   Limited out-ICU data\n",
    "*   \\>60,000 ICU stays and \\>40,000 patients\n",
    "*   **Real Data**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Yt6Em17Sw8R4"
   },
   "source": [
    "**Accessing the data**\n",
    "\n",
    "\n",
    "*   *Colab*: Jupyter notebook environment, run queries, gather and process data.\n",
    "*   [*BigQuery*](https://bigquery.cloud.google.com/): Run queries and explore data.\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GBVYck3T8Rjs"
   },
   "source": [
    "**Setting up Colab**\n",
    "\n",
    "To query data using BigQuery you'll need to import some libraries. Just run the following cell:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "lyi5zSjq9mTa"
   },
   "outputs": [],
   "source": [
    "# Imports required to access datathon data\n",
    "from google.colab import auth\n",
    "from google.cloud import bigquery\n",
    "import os\n",
    "\n",
    "# Feel free to import other libraries\n",
    "import tensorflow as tf\n",
    "import seaborn as sns\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "oTVFUnT49_4n"
   },
   "source": [
    "Authenticate yourself, follow the link to login with your gmail if requested and copy/paste the verification code:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "2L0fd7Yq-oZu"
   },
   "outputs": [],
   "source": [
    "auth.authenticate_user()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jZtKdUEO_D98"
   },
   "source": [
    "Finally link the notebook to BigQuery:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ikguyJ9D_btr"
   },
   "outputs": [],
   "source": [
    "project_id='hack-aotearoa'\n",
    "os.environ[\"GOOGLE_CLOUD_PROJECT\"]=project_id\n",
    "\n",
    "\n",
    "# Returns a dataframe given query string as input. \n",
    "def run_query(query):\n",
    "  return pd.io.gbq.read_gbq(query, project_id=project_id, configuration={'query':{'useLegacySql': False}}, dialect = 'legacy')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "2ngcGjGaAp_T"
   },
   "source": [
    "Let's run a query to show how to pull data:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 359
    },
    "colab_type": "code",
    "id": "uMxeW_K_Avo_",
    "outputId": "874504c0-4954-4ae0-fed5-b7a930e9f0dd"
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ROW_ID</th>\n",
       "      <th>SUBJECT_ID</th>\n",
       "      <th>GENDER</th>\n",
       "      <th>DOB</th>\n",
       "      <th>DOD</th>\n",
       "      <th>DOD_HOSP</th>\n",
       "      <th>DOD_SSN</th>\n",
       "      <th>EXPIRE_FLAG</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>234</td>\n",
       "      <td>249</td>\n",
       "      <td>F</td>\n",
       "      <td>2075-03-13</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>238</td>\n",
       "      <td>253</td>\n",
       "      <td>F</td>\n",
       "      <td>2089-11-26</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>242</td>\n",
       "      <td>258</td>\n",
       "      <td>F</td>\n",
       "      <td>2124-09-19</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>243</td>\n",
       "      <td>260</td>\n",
       "      <td>F</td>\n",
       "      <td>2105-03-23</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>247</td>\n",
       "      <td>264</td>\n",
       "      <td>F</td>\n",
       "      <td>2162-11-30</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>249</td>\n",
       "      <td>266</td>\n",
       "      <td>F</td>\n",
       "      <td>2090-12-17</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>250</td>\n",
       "      <td>267</td>\n",
       "      <td>F</td>\n",
       "      <td>2131-09-05</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>626</td>\n",
       "      <td>663</td>\n",
       "      <td>F</td>\n",
       "      <td>2086-05-12</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>630</td>\n",
       "      <td>667</td>\n",
       "      <td>F</td>\n",
       "      <td>2053-09-17</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>638</td>\n",
       "      <td>675</td>\n",
       "      <td>F</td>\n",
       "      <td>2158-04-05</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>None</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   ROW_ID  SUBJECT_ID GENDER        DOB   DOD DOD_HOSP DOD_SSN  EXPIRE_FLAG\n",
       "0     234         249      F 2075-03-13  None     None    None            0\n",
       "1     238         253      F 2089-11-26  None     None    None            0\n",
       "2     242         258      F 2124-09-19  None     None    None            0\n",
       "3     243         260      F 2105-03-23  None     None    None            0\n",
       "4     247         264      F 2162-11-30  None     None    None            0\n",
       "5     249         266      F 2090-12-17  None     None    None            0\n",
       "6     250         267      F 2131-09-05  None     None    None            0\n",
       "7     626         663      F 2086-05-12  None     None    None            0\n",
       "8     630         667      F 2053-09-17  None     None    None            0\n",
       "9     638         675      F 2158-04-05  None     None    None            0"
      ]
     },
     "execution_count": 15,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "example_query = '''\n",
    "SELECT *\n",
    "FROM `physionet-data.mimiciii_clinical.patients` p\n",
    "LIMIT 100\n",
    "'''\n",
    "\n",
    "run_query(example_query).head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "gmXDERgeDU75"
   },
   "source": [
    "Now let's try to compute the age at admission to the ICU using a simple join:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 286
    },
    "colab_type": "code",
    "id": "z2fe0n_PDg6B",
    "outputId": "370bb1b8-9021-4eb5-8b13-c17fc99428f2"
   },
   "outputs": [],
   "source": [
    "age_query = '''\n",
    "SELECT i.icustay_id, p.gender, DATETIME_DIFF(i.intime, p.dob, YEAR) AS age \n",
    "FROM `physionet-data.mimiciii_clinical.icustays` i\n",
    "INNER JOIN `physionet-data.mimiciii_clinical.patients` p\n",
    "ON i.subject_id = p.subject_id'''\n",
    "\n",
    "age_df = run_query(age_query)\n",
    "age_df.age.hist(bins = np.arange(18, 310, 1))\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "HJ_VStPPEjQR"
   },
   "source": [
    "Note that dates for patients older than 89 at admission were manipulated so to show them to be older than 300 for privacy reasons."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "oIjLaQbp_2y0"
   },
   "source": [
    "*You are good to go*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "xMmPm3xl32zP"
   },
   "source": [
    "# **Predicting septic patients in-hospital mortality with physiological data**\n",
    "\n",
    "\n",
    "Sepsis is a life threatening condition which requires prompt intervention from physicians and may lead to tissue damage, organ failure and potentially death.\n",
    "It is estimated that 30 milion people worldwide are affected by sepsis and ,of those, 6 milions die from it. The incidence of sepsis rose significantly in the past decade and, in the U.S, accounts for 13% of healthcare expenditures[1][2].\n",
    "Despite the relevance of Sepsis, an absolute consensus on its defintion has never been fully reached from the international community and several identification methods were proposed during the years, the most recent being Sepsis-3[3].\n",
    "\n",
    "The goal of this study is to assess whether it is possible to predict mortality of patients with angus sepsis using commonly measured vitals, scales and lab results in the first 48 hours of ICU from the MIMIC database.\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Pz01KmgYESE0"
   },
   "source": [
    "**A: Cohort Definiton**\n",
    "\n",
    "First, we want to define the cohort of patients. \n",
    "Particularly we are interested in those patients that:\n",
    "\n",
    "\n",
    "*   Developed Angus sepsis.\n",
    "*   Have age between 18 and 89.\n",
    "*   Have stayed the ICU more than 48 hours.\n",
    "\n",
    "At this point we may want to also extract features such as age, gender and in-hospital mortality, the latter one will be the outcome used to train our model. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 359
    },
    "colab_type": "code",
    "id": "rbK1W_6yBWEY",
    "outputId": "e158593a-a5c5-49d4-89f7-86c43353248b"
   },
   "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>icustay_id</th>\n",
       "      <th>gender</th>\n",
       "      <th>admission_age</th>\n",
       "      <th>hospital_expire_flag</th>\n",
       "      <th>intime</th>\n",
       "      <th>icustay_seq</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>1</td>\n",
       "      <td>88</td>\n",
       "      <td>1</td>\n",
       "      <td>2135-01-30 20:53:34</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>38</td>\n",
       "      <td>185910</td>\n",
       "      <td>248910</td>\n",
       "      <td>1</td>\n",
       "      <td>76</td>\n",
       "      <td>0</td>\n",
       "      <td>2166-08-10 00:29:36</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>117</td>\n",
       "      <td>164853</td>\n",
       "      <td>217966</td>\n",
       "      <td>0</td>\n",
       "      <td>50</td>\n",
       "      <td>1</td>\n",
       "      <td>2133-11-18 13:24:19</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>124</td>\n",
       "      <td>138376</td>\n",
       "      <td>256064</td>\n",
       "      <td>1</td>\n",
       "      <td>76</td>\n",
       "      <td>1</td>\n",
       "      <td>2166-01-09 13:05:18</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>157</td>\n",
       "      <td>107880</td>\n",
       "      <td>264885</td>\n",
       "      <td>1</td>\n",
       "      <td>81</td>\n",
       "      <td>0</td>\n",
       "      <td>2106-06-17 19:51:46</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>191</td>\n",
       "      <td>136614</td>\n",
       "      <td>280149</td>\n",
       "      <td>1</td>\n",
       "      <td>73</td>\n",
       "      <td>0</td>\n",
       "      <td>2196-04-09 19:52:59</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>195</td>\n",
       "      <td>118936</td>\n",
       "      <td>291941</td>\n",
       "      <td>0</td>\n",
       "      <td>74</td>\n",
       "      <td>0</td>\n",
       "      <td>2167-12-02 13:16:56</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>195</td>\n",
       "      <td>118936</td>\n",
       "      <td>266038</td>\n",
       "      <td>0</td>\n",
       "      <td>74</td>\n",
       "      <td>0</td>\n",
       "      <td>2167-12-11 12:49:33</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>211</td>\n",
       "      <td>101148</td>\n",
       "      <td>248590</td>\n",
       "      <td>0</td>\n",
       "      <td>85</td>\n",
       "      <td>0</td>\n",
       "      <td>2197-12-17 02:53:02</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>223</td>\n",
       "      <td>105694</td>\n",
       "      <td>255617</td>\n",
       "      <td>1</td>\n",
       "      <td>68</td>\n",
       "      <td>0</td>\n",
       "      <td>2157-05-01 05:20:41</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   subject_id  hadm_id  ...              intime  icustay_seq\n",
       "0          21   111970  ... 2135-01-30 20:53:34            1\n",
       "1          38   185910  ... 2166-08-10 00:29:36            1\n",
       "2         117   164853  ... 2133-11-18 13:24:19            1\n",
       "3         124   138376  ... 2166-01-09 13:05:18            1\n",
       "4         157   107880  ... 2106-06-17 19:51:46            1\n",
       "5         191   136614  ... 2196-04-09 19:52:59            1\n",
       "6         195   118936  ... 2167-12-02 13:16:56            1\n",
       "7         195   118936  ... 2167-12-11 12:49:33            2\n",
       "8         211   101148  ... 2197-12-17 02:53:02            1\n",
       "9         223   105694  ... 2157-05-01 05:20:41            1\n",
       "\n",
       "[10 rows x 8 columns]"
      ]
     },
     "execution_count": 31,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# hint: use the icustay_detail and angus_sepsis tables in the mimiciii_derived database\n",
    "\n",
    "cohort_query = '''\n",
    "SELECT i.subject_id, i.hadm_id, i.icustay_id, i.gender, i.admission_age, i.hospital_expire_flag, i.intime, i.icustay_seq\n",
    "FROM `physionet-data.mimiciii_derived.icustay_detail` i\n",
    "INNER JOIN `physionet-data.mimiciii_derived.angus_sepsis` a\n",
    "ON a.hadm_id = i.hadm_id AND a.explicit_sepsis = 1\n",
    "WHERE i.admission_age >= 18 AND i.admission_age < 89 AND i.los_icu >= 2\n",
    "'''\n",
    "\n",
    "\n",
    "cohort = run_query(cohort_query)\n",
    "\n",
    "cohort = cohort.replace({'gender': {'F': 0, \n",
    "                                    'M': 1}})\n",
    "\n",
    "\n",
    "cohort.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "tuAnZ8RwH_yc"
   },
   "source": [
    "It is good practice to run some statistics to have an overview of the cohort:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "id": "6Vy9KfRo8hs0",
    "outputId": "ba521325-51da-4d88-dba3-997927f0893f"
   },
   "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 tr th {\n",
       "        text-align: left;\n",
       "    }\n",
       "\n",
       "    .dataframe thead tr:last-of-type th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>icustay_id</th>\n",
       "      <th>gender</th>\n",
       "      <th colspan=\"2\" halign=\"left\">admission_age</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th></th>\n",
       "      <th>nunique</th>\n",
       "      <th>mean</th>\n",
       "      <th>mean</th>\n",
       "      <th>std</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hospital_expire_flag</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2282.0</td>\n",
       "      <td>0.564417</td>\n",
       "      <td>63.906661</td>\n",
       "      <td>15.177008</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1324.0</td>\n",
       "      <td>0.597432</td>\n",
       "      <td>66.434290</td>\n",
       "      <td>14.209354</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>tot</th>\n",
       "      <td>3606.0</td>\n",
       "      <td>0.576539</td>\n",
       "      <td>64.834720</td>\n",
       "      <td>14.877060</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     icustay_id    gender admission_age           \n",
       "                        nunique      mean          mean        std\n",
       "hospital_expire_flag                                              \n",
       "0                        2282.0  0.564417     63.906661  15.177008\n",
       "1                        1324.0  0.597432     66.434290  14.209354\n",
       "tot                      3606.0  0.576539     64.834720  14.877060"
      ]
     },
     "execution_count": 54,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cohort_stats = cohort.groupby('hospital_expire_flag').agg({'icustay_id': 'nunique', 'gender': 'mean', 'admission_age': ['mean', 'std']})\n",
    "cohort_stats.loc['tot'] = cohort.agg({'icustay_id': 'nunique', 'gender': 'mean', 'admission_age': ['mean', 'std']}).stack().swaplevel()\n",
    "\n",
    "cohort_stats"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TJVEfm0BRu4j"
   },
   "source": [
    "**B: Querying vitals**\n",
    "\n",
    "Most charted observations can be found in the chartevents table, like the vital parameters we need. The chartevents table is the largest in the MIMIC and contains many important variables like blood pressure readings, heart rate and respiration. \n",
    "We are now interested in:\n",
    "\n",
    "\n",
    "*   Systolic blood pressure\n",
    "*   Diastolic blood pressure\n",
    "*   Heart rate \n",
    "*   Respiratory rate \n",
    "*   Temperature\n",
    "*   Oxygen saturation (SpO2)\n",
    "\n",
    "We want to query these variables for all patients defined in the cohort, but only for the first 48 hours. Be very carefull with units, as temperature for example is measured in both Celsius and Fahrenheit."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 359
    },
    "colab_type": "code",
    "id": "Ur1QQsuf0hIv",
    "outputId": "9da2d594-a28e-4a73-a14f-79f327869f63"
   },
   "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>icustay_id</th>\n",
       "      <th>charttime</th>\n",
       "      <th>delta</th>\n",
       "      <th>valuenum</th>\n",
       "      <th>type</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>1.116667</td>\n",
       "      <td>79.000000</td>\n",
       "      <td>HR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>1.116667</td>\n",
       "      <td>60.000000</td>\n",
       "      <td>DBP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>1.116667</td>\n",
       "      <td>114.000000</td>\n",
       "      <td>SBP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>1.116667</td>\n",
       "      <td>37.611099</td>\n",
       "      <td>TMP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>1.116667</td>\n",
       "      <td>37.611109</td>\n",
       "      <td>TMP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>1.116667</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>RSP</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>1.116667</td>\n",
       "      <td>98.000000</td>\n",
       "      <td>SpO2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 23:00:00</td>\n",
       "      <td>2.116667</td>\n",
       "      <td>100.000000</td>\n",
       "      <td>SpO2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 23:00:00</td>\n",
       "      <td>2.116667</td>\n",
       "      <td>75.000000</td>\n",
       "      <td>HR</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 23:00:00</td>\n",
       "      <td>2.116667</td>\n",
       "      <td>82.000000</td>\n",
       "      <td>SBP</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   subject_id  hadm_id  icustay_id  ...     delta    valuenum  type\n",
       "0          21   111970      216859  ...  1.116667   79.000000    HR\n",
       "1          21   111970      216859  ...  1.116667   60.000000   DBP\n",
       "2          21   111970      216859  ...  1.116667  114.000000   SBP\n",
       "3          21   111970      216859  ...  1.116667   37.611099   TMP\n",
       "4          21   111970      216859  ...  1.116667   37.611109   TMP\n",
       "5          21   111970      216859  ...  1.116667   12.000000   RSP\n",
       "6          21   111970      216859  ...  1.116667   98.000000  SpO2\n",
       "7          21   111970      216859  ...  2.116667  100.000000  SpO2\n",
       "8          21   111970      216859  ...  2.116667   75.000000    HR\n",
       "9          21   111970      216859  ...  2.116667   82.000000   SBP\n",
       "\n",
       "[10 rows x 7 columns]"
      ]
     },
     "execution_count": 55,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Use the table `physionet-data.mimiciii_clinical.chartevents`\n",
    "# hint: use table d_items and SQL function LIKE to find variables.\n",
    "# hint: to keep only observations from the first 48 hours use the function DATETIME_DIFF with ICU intime.\n",
    "\n",
    "\n",
    "vitals_query = '''\n",
    "WITH cohort AS(\n",
    "SELECT i.subject_id, i.hadm_id, i.icustay_id, i.gender, i.admission_age, i.hospital_expire_flag, i.intime\n",
    "FROM `physionet-data.mimiciii_derived.icustay_detail` i\n",
    "INNER JOIN `physionet-data.mimiciii_derived.angus_sepsis` a\n",
    "ON a.hadm_id = i.hadm_id AND a.explicit_sepsis = 1\n",
    "WHERE i.admission_age >= 18 AND i.admission_age < 89 AND i.los_icu >= 2\n",
    ")\n",
    ", vitals AS(\n",
    "SELECT c.subject_id, c.hadm_id, c.icustay_id, c.charttime, DATETIME_DIFF(c.charttime, i.intime, MINUTE)/60 AS delta, \n",
    "CASE \n",
    "  WHEN c.itemid IN (223761, 678, 679) THEN (c.valuenum - 32)*(5/9) ELSE c.valuenum END AS valuenum, \n",
    "CASE \n",
    "\tWHEN c.itemid IN (51, 442, 455, 6701, 220179, 220050)         THEN 'SBP'\n",
    "  WHEN c.itemid IN (8368, 8440, 8441, 8555, 220180, 220051)     THEN 'DBP'\n",
    "  WHEN c.itemid IN (618, 615, 220210, 224690)                   THEN 'RSP'\n",
    "  WHEN c.itemid IN (223762, 676, 223761, 678, 677, 679)         THEN 'TMP'\n",
    "  WHEN c.itemid IN (211, 220045)                                THEN 'HR'\n",
    "  WHEN c.itemid IN (646, 22027)                                 THEN 'SpO2'\n",
    "  ELSE NULL\n",
    "  END AS type\n",
    "FROM `physionet-data.mimiciii_clinical.chartevents` c\n",
    "INNER JOIN cohort i\n",
    "ON i.icustay_id = c.icustay_id \n",
    "WHERE DATETIME_DIFF(c.charttime, i.intime, MINUTE)/60 <= 48\n",
    "ORDER BY c.subject_id, c.hadm_id, c.icustay_id, c.charttime)\n",
    "SELECT *\n",
    "FROM vitals v\n",
    "WHERE v.type IS NOT NULL\n",
    "'''\n",
    "\n",
    "vitals = run_query(vitals_query)\n",
    "vitals.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5K0dI4hapdVi"
   },
   "source": [
    "How to add Mean Arterial Pressure (MAP) to the above query?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "htRMtIGxTbf9"
   },
   "source": [
    "**C: Lab results**\n",
    "\n",
    "Lab test are another source with which it is possible to the describe physiological state of patients. Similarly for vitals, we want to extract all variables form the pivoted_lab table in the MIMIC derived database. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 529
    },
    "colab_type": "code",
    "id": "mDXRhRnW51dt",
    "outputId": "de9189cb-af7a-4cfd-9118-2414c373dd73"
   },
   "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>charttime</th>\n",
       "      <th>ANIONGAP</th>\n",
       "      <th>ALBUMIN</th>\n",
       "      <th>BANDS</th>\n",
       "      <th>BICARBONATE</th>\n",
       "      <th>BILIRUBIN</th>\n",
       "      <th>CREATININE</th>\n",
       "      <th>CHLORIDE</th>\n",
       "      <th>GLUCOSE</th>\n",
       "      <th>HEMATOCRIT</th>\n",
       "      <th>HEMOGLOBIN</th>\n",
       "      <th>LACTATE</th>\n",
       "      <th>PLATELET</th>\n",
       "      <th>POTASSIUM</th>\n",
       "      <th>PTT</th>\n",
       "      <th>INR</th>\n",
       "      <th>PT</th>\n",
       "      <th>SODIUM</th>\n",
       "      <th>BUN</th>\n",
       "      <th>WBC</th>\n",
       "      <th>delta</th>\n",
       "      <th>icustay_id</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>2135-01-30 19:05:00</td>\n",
       "      <td>19.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.0</td>\n",
       "      <td>23.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.2</td>\n",
       "      <td>99.0</td>\n",
       "      <td>200.0</td>\n",
       "      <td>29.6</td>\n",
       "      <td>9.1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>275.0</td>\n",
       "      <td>3.8</td>\n",
       "      <td>31.7</td>\n",
       "      <td>1.7</td>\n",
       "      <td>18.1</td>\n",
       "      <td>137.0</td>\n",
       "      <td>49.0</td>\n",
       "      <td>30.5</td>\n",
       "      <td>-1.800000</td>\n",
       "      <td>216859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>2135-01-30 19:18:00</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>NaN</td>\n",
       "      <td>1.6</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>-1.583333</td>\n",
       "      <td>216859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>2135-01-31 05:20:00</td>\n",
       "      <td>15.0</td>\n",
       "      <td>2.1</td>\n",
       "      <td>NaN</td>\n",
       "      <td>24.0</td>\n",
       "      <td>0.3</td>\n",
       "      <td>4.2</td>\n",
       "      <td>104.0</td>\n",
       "      <td>178.0</td>\n",
       "      <td>30.6</td>\n",
       "      <td>9.3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>432.0</td>\n",
       "      <td>3.4</td>\n",
       "      <td>43.5</td>\n",
       "      <td>1.8</td>\n",
       "      <td>19.1</td>\n",
       "      <td>140.0</td>\n",
       "      <td>50.0</td>\n",
       "      <td>41.2</td>\n",
       "      <td>8.450000</td>\n",
       "      <td>216859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>2135-02-01 02:59:00</td>\n",
       "      <td>19.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>20.0</td>\n",
       "      <td>0.2</td>\n",
       "      <td>4.7</td>\n",
       "      <td>105.0</td>\n",
       "      <td>178.0</td>\n",
       "      <td>33.5</td>\n",
       "      <td>9.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>387.0</td>\n",
       "      <td>4.6</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>139.0</td>\n",
       "      <td>63.0</td>\n",
       "      <td>38.6</td>\n",
       "      <td>30.100000</td>\n",
       "      <td>216859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>2135-02-01 12:42:00</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>NaN</td>\n",
       "      <td>2.7</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>39.816667</td>\n",
       "      <td>216859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>2135-02-01 18:00:00</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>NaN</td>\n",
       "      <td>1.6</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>45.116667</td>\n",
       "      <td>216859</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>38</td>\n",
       "      <td>185910</td>\n",
       "      <td>2166-08-10 00:17:00</td>\n",
       "      <td>17.0</td>\n",
       "      <td>2.7</td>\n",
       "      <td>NaN</td>\n",
       "      <td>21.0</td>\n",
       "      <td>1.3</td>\n",
       "      <td>1.8</td>\n",
       "      <td>106.0</td>\n",
       "      <td>94.0</td>\n",
       "      <td>36.0</td>\n",
       "      <td>12.5</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.0</td>\n",
       "      <td>4.4</td>\n",
       "      <td>28.2</td>\n",
       "      <td>1.4</td>\n",
       "      <td>14.4</td>\n",
       "      <td>140.0</td>\n",
       "      <td>42.0</td>\n",
       "      <td>18.2</td>\n",
       "      <td>-0.200000</td>\n",
       "      <td>248910</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>38</td>\n",
       "      <td>185910</td>\n",
       "      <td>2166-08-10 16:13:00</td>\n",
       "      <td>16.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>20.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.6</td>\n",
       "      <td>107.0</td>\n",
       "      <td>137.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>4.2</td>\n",
       "      <td>50.6</td>\n",
       "      <td>1.7</td>\n",
       "      <td>15.6</td>\n",
       "      <td>139.0</td>\n",
       "      <td>55.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>15.733333</td>\n",
       "      <td>248910</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>38</td>\n",
       "      <td>185910</td>\n",
       "      <td>2166-08-11 00:01:00</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>NaN</td>\n",
       "      <td>2.8</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>23.533333</td>\n",
       "      <td>248910</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>38</td>\n",
       "      <td>185910</td>\n",
       "      <td>2166-08-11 01:30:00</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>53.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>25.016667</td>\n",
       "      <td>248910</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   subject_id  hadm_id           charttime  ...   WBC      delta  icustay_id\n",
       "0          21   111970 2135-01-30 19:05:00  ...  30.5  -1.800000      216859\n",
       "1          21   111970 2135-01-30 19:18:00  ...   NaN  -1.583333      216859\n",
       "2          21   111970 2135-01-31 05:20:00  ...  41.2   8.450000      216859\n",
       "3          21   111970 2135-02-01 02:59:00  ...  38.6  30.100000      216859\n",
       "4          21   111970 2135-02-01 12:42:00  ...   NaN  39.816667      216859\n",
       "5          21   111970 2135-02-01 18:00:00  ...   NaN  45.116667      216859\n",
       "6          38   185910 2166-08-10 00:17:00  ...  18.2  -0.200000      248910\n",
       "7          38   185910 2166-08-10 16:13:00  ...   NaN  15.733333      248910\n",
       "8          38   185910 2166-08-11 00:01:00  ...   NaN  23.533333      248910\n",
       "9          38   185910 2166-08-11 01:30:00  ...   NaN  25.016667      248910\n",
       "\n",
       "[10 rows x 24 columns]"
      ]
     },
     "execution_count": 56,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# hint: lab tests are not associated to the icustay_id but to the hadm_id, filter values based on ICU admission time (intime)\n",
    "\n",
    "lab_query = '''\n",
    "WITH cohort AS(\n",
    "SELECT i.subject_id, i.hadm_id, i.icustay_id, i.gender, i.admission_age, i.hospital_expire_flag, i.intime\n",
    "FROM `physionet-data.mimiciii_derived.icustay_detail` i\n",
    "INNER JOIN `physionet-data.mimiciii_derived.angus_sepsis` a\n",
    "ON a.hadm_id = i.hadm_id AND a.explicit_sepsis = 1\n",
    "WHERE i.admission_age >= 18 AND i.admission_age < 89 AND i.los_icu >= 2\n",
    ")\n",
    "SELECT l.*, DATETIME_DIFF(l.charttime, i.intime, MINUTE)/60 AS delta, i.icustay_id\n",
    "FROM `physionet-data.mimiciii_derived.pivoted_lab` l\n",
    "INNER JOIN cohort i \n",
    "ON i.hadm_id = l.hadm_id \n",
    "WHERE DATETIME_DIFF(l.charttime, i.intime, MINUTE)/60 >= -6 AND DATETIME_DIFF(l.charttime, i.intime, MINUTE)/60 <= 48\n",
    "ORDER BY l.subject_id, l.hadm_id, i.icustay_id, l.charttime\n",
    "'''\n",
    "\n",
    "lab = run_query(lab_query)\n",
    "lab.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "RguN6k2qpyOB"
   },
   "source": [
    "How often are lab values measured? \n",
    "does the frequency of measurement correlate with the outcome?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "cxYGBqxzUfUX"
   },
   "source": [
    "**D: Glasgow Coma Scale**\n",
    "\n",
    "The Galsgow Coma Scale (GCS) provides the state of a person's consciousness. The GCS is the combination of three indicators:\n",
    "\n",
    "\n",
    "*   GCS Motor\n",
    "*   GCS Verbal\n",
    "*   GCS Eyes\n",
    "\n",
    "We may query those too.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 359
    },
    "colab_type": "code",
    "id": "6me0XEL56CWu",
    "outputId": "685e4d0e-00f2-4db7-aeca-2a75d578847d"
   },
   "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>icustay_id</th>\n",
       "      <th>charttime</th>\n",
       "      <th>GCS</th>\n",
       "      <th>GCSMotor</th>\n",
       "      <th>GCSVerbal</th>\n",
       "      <th>GCSEyes</th>\n",
       "      <th>EndoTrachFlag</th>\n",
       "      <th>subject_id</th>\n",
       "      <th>hadm_id</th>\n",
       "      <th>delta</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-30 22:00:00</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>1.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-31 06:00:00</td>\n",
       "      <td>13.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>9.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-31 08:00:00</td>\n",
       "      <td>13.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>11.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-31 10:30:00</td>\n",
       "      <td>13.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>13.616667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-31 12:00:00</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>15.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-31 16:00:00</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>19.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-01-31 20:00:00</td>\n",
       "      <td>14.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>23.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-02-01 00:00:00</td>\n",
       "      <td>15.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>27.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-02-01 04:00:00</td>\n",
       "      <td>15.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>31.116667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>216859</td>\n",
       "      <td>2135-02-01 08:00:00</td>\n",
       "      <td>15.0</td>\n",
       "      <td>6.0</td>\n",
       "      <td>5.0</td>\n",
       "      <td>4.0</td>\n",
       "      <td>0</td>\n",
       "      <td>21</td>\n",
       "      <td>111970</td>\n",
       "      <td>35.116667</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   icustay_id           charttime   GCS  ...  subject_id  hadm_id      delta\n",
       "0      216859 2135-01-30 22:00:00  14.0  ...          21   111970   1.116667\n",
       "1      216859 2135-01-31 06:00:00  13.0  ...          21   111970   9.116667\n",
       "2      216859 2135-01-31 08:00:00  13.0  ...          21   111970  11.116667\n",
       "3      216859 2135-01-31 10:30:00  13.0  ...          21   111970  13.616667\n",
       "4      216859 2135-01-31 12:00:00  14.0  ...          21   111970  15.116667\n",
       "5      216859 2135-01-31 16:00:00  14.0  ...          21   111970  19.116667\n",
       "6      216859 2135-01-31 20:00:00  14.0  ...          21   111970  23.116667\n",
       "7      216859 2135-02-01 00:00:00  15.0  ...          21   111970  27.116667\n",
       "8      216859 2135-02-01 04:00:00  15.0  ...          21   111970  31.116667\n",
       "9      216859 2135-02-01 08:00:00  15.0  ...          21   111970  35.116667\n",
       "\n",
       "[10 rows x 10 columns]"
      ]
     },
     "execution_count": 57,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# hint: use the pivoted_gcs in mimiciii_derived\n",
    "\n",
    "gcs_query = '''\n",
    "WITH cohort AS(\n",
    "SELECT i.subject_id, i.hadm_id, i.icustay_id, i.gender, i.admission_age, i.hospital_expire_flag, i.intime\n",
    "FROM `physionet-data.mimiciii_derived.icustay_detail` i\n",
    "INNER JOIN `physionet-data.mimiciii_derived.angus_sepsis` a\n",
    "ON a.hadm_id = i.hadm_id AND a.explicit_sepsis = 1\n",
    "WHERE i.admission_age >= 18 AND i.admission_age < 89 AND i.los_icu >= 2\n",
    ")\n",
    "SELECT l.*, i.subject_id, i.hadm_id, DATETIME_DIFF(l.charttime, i.intime, MINUTE)/60 AS delta\n",
    "FROM `physionet-data.mimiciii_derived.pivoted_gcs` l\n",
    "INNER JOIN cohort i \n",
    "ON i.icustay_id = l.icustay_id \n",
    "WHERE DATETIME_DIFF(l.charttime, i.intime, MINUTE)/60 >= -6 AND DATETIME_DIFF(l.charttime, i.intime, MINUTE)/60 <= 48\n",
    "ORDER BY i.subject_id, i.hadm_id, i.icustay_id, l.charttime'''\n",
    "\n",
    "gcs = run_query(gcs_query)\n",
    "gcs.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "zygsy4_4VkKL"
   },
   "source": [
    "**E: Building the dataset**\n",
    "\n",
    "Vitals data are structured differently from lab and gcs data (both pivoted). We want to merge everything together so that the end result matains the structure of gcs and lab. One approach to do this is to first stack the pivoted tables, gcs and lab. Now, vitals, gcs and lab share the same format. Leveraging pandas, we can use dataframes concatenation with pivoting to obtain the final dataset. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 379
    },
    "colab_type": "code",
    "id": "CkLoynK2-N-m",
    "outputId": "6f6cf445-a6db-4c73-92c5-491d9e753736"
   },
   "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>type</th>\n",
       "      <th>icustay_id</th>\n",
       "      <th>delta</th>\n",
       "      <th>ALBUMIN</th>\n",
       "      <th>ANIONGAP</th>\n",
       "      <th>BANDS</th>\n",
       "      <th>BICARBONATE</th>\n",
       "      <th>BILIRUBIN</th>\n",
       "      <th>BUN</th>\n",
       "      <th>CHLORIDE</th>\n",
       "      <th>CREATININE</th>\n",
       "      <th>DBP</th>\n",
       "      <th>EndoTrachFlag</th>\n",
       "      <th>GCS</th>\n",
       "      <th>GCSEyes</th>\n",
       "      <th>GCSMotor</th>\n",
       "      <th>GCSVerbal</th>\n",
       "      <th>GLUCOSE</th>\n",
       "      <th>HEMATOCRIT</th>\n",
       "      <th>HEMOGLOBIN</th>\n",
       "      <th>HR</th>\n",
       "      <th>INR</th>\n",
       "      <th>LACTATE</th>\n",
       "      <th>PLATELET</th>\n",
       "      <th>POTASSIUM</th>\n",
       "      <th>PT</th>\n",
       "      <th>PTT</th>\n",
       "      <th>RSP</th>\n",
       "      <th>SBP</th>\n",
       "      <th>SODIUM</th>\n",
       "      <th>SpO2</th>\n",
       "      <th>TMP</th>\n",
       "      <th>WBC</th>\n",
       "      <th>MAP</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>200030</td>\n",
       "      <td>-5.166667</td>\n",
       "      <td>NaN</td>\n",
       "      <td>25.0</td>\n",
       "      <td>8.0</td>\n",
       "      <td>16.0</td>\n",
       "      <td>3.6</td>\n",
       "      <td>40.0</td>\n",
       "      <td>93.0</td>\n",
       "      <td>1.9</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>524.0</td>\n",
       "      <td>34.5</td>\n",
       "      <td>11.4</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>142.0</td>\n",
       "      <td>4.2</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>130.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>9.4</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>200030</td>\n",
       "      <td>-4.983333</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>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>NaN</td>\n",
       "      <td>6.8</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>200030</td>\n",
       "      <td>-4.650000</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>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>NaN</td>\n",
       "      <td>5.8</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>200030</td>\n",
       "      <td>-3.700000</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>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>1.9</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>16.6</td>\n",
       "      <td>28.3</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",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>200030</td>\n",
       "      <td>-2.466667</td>\n",
       "      <td>NaN</td>\n",
       "      <td>21.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>15.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>39.0</td>\n",
       "      <td>98.0</td>\n",
       "      <td>1.8</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>403.0</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>4.3</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>130.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>200030</td>\n",
       "      <td>-2.300000</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>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>NaN</td>\n",
       "      <td>6.2</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>200030</td>\n",
       "      <td>-0.883333</td>\n",
       "      <td>NaN</td>\n",
       "      <td>17.0</td>\n",
       "      <td>3.0</td>\n",
       "      <td>15.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>36.0</td>\n",
       "      <td>104.0</td>\n",
       "      <td>1.4</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>363.0</td>\n",
       "      <td>29.6</td>\n",
       "      <td>9.8</td>\n",
       "      <td>NaN</td>\n",
       "      <td>2.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>139.0</td>\n",
       "      <td>3.4</td>\n",
       "      <td>16.9</td>\n",
       "      <td>30.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>133.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>11.7</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>200030</td>\n",
       "      <td>-0.450000</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>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>NaN</td>\n",
       "      <td>4.4</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>NaN</td>\n",
       "      <td>NaN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>200030</td>\n",
       "      <td>-0.383333</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>65.0</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>83.0</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>123.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>92.0</td>\n",
       "      <td>36.111105</td>\n",
       "      <td>NaN</td>\n",
       "      <td>84.333333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>200030</td>\n",
       "      <td>-0.133333</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>65.0</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>84.0</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>21.0</td>\n",
       "      <td>98.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>93.0</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>76.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "type  icustay_id     delta  ALBUMIN  ANIONGAP  ...  SpO2        TMP   WBC        MAP\n",
       "0         200030 -5.166667      NaN      25.0  ...   NaN        NaN   9.4        NaN\n",
       "1         200030 -4.983333      NaN       NaN  ...   NaN        NaN   NaN        NaN\n",
       "2         200030 -4.650000      NaN       NaN  ...   NaN        NaN   NaN        NaN\n",
       "3         200030 -3.700000      NaN       NaN  ...   NaN        NaN   NaN        NaN\n",
       "4         200030 -2.466667      NaN      21.0  ...   NaN        NaN   NaN        NaN\n",
       "5         200030 -2.300000      NaN       NaN  ...   NaN        NaN   NaN        NaN\n",
       "6         200030 -0.883333      NaN      17.0  ...   NaN        NaN  11.7        NaN\n",
       "7         200030 -0.450000      NaN       NaN  ...   NaN        NaN   NaN        NaN\n",
       "8         200030 -0.383333      NaN       NaN  ...  92.0  36.111105   NaN  84.333333\n",
       "9         200030 -0.133333      NaN       NaN  ...  93.0        NaN   NaN  76.000000\n",
       "\n",
       "[10 rows x 33 columns]"
      ]
     },
     "execution_count": 59,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "p_gcs = gcs.set_index(['subject_id','hadm_id','icustay_id','charttime','delta']).stack().reset_index().rename({0: 'valuenum', 'level_5': 'type'}, axis = 1) # stacking the gcs table \n",
    "p_lab = lab.set_index(['subject_id','hadm_id','icustay_id','charttime','delta']).stack().reset_index().rename({0: 'valuenum', 'level_5': 'type'}, axis = 1) # stacking lab \n",
    "\n",
    "t0 = pd.concat([vitals, p_gcs, p_lab], axis = 0, sort = True) # now that gcs and lab have the same format as vitals, we can concatenate the theree tables along the rows.\n",
    "ts_df = pd.pivot_table(t0, index = ['icustay_id', 'delta'], values = 'valuenum', columns = 'type') # applying the pivot operator.\n",
    "ts_df = ts_df.reset_index()\n",
    "ts_df['MAP'] = (ts_df['SBP'] + 2*ts_df['DBP'])/3 # combining features in this format is far easier, for example we can compute the mean arterial pressure easily.\n",
    "ts_df.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mGYuacm4WenQ"
   },
   "source": [
    "We can verify the results by plotting some timeseries for a patient."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 300
    },
    "colab_type": "code",
    "id": "vTtz4c2kAAOs",
    "outputId": "f82fcf20-5e7b-479b-f4e2-41503b2f5b8b"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f1ecd7cd9e8>"
      ]
     },
     "execution_count": 60,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "ts_df.loc[ts_df.icustay_id == 200030, ['delta','MAP', 'HR', 'RSP','TMP','SpO2']].plot(x = 'delta', grid = {'alpha': .3})"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "ZOJWbo2sXA0X"
   },
   "source": [
    "**F: Feature engineering**\n",
    "\n",
    "To begin, let's keep it simple and just compute the mean and standard deviation of every variable as features. After computing the mean we should add the outcome (hospital_expire_flag), age and gender back to out dataset."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 461
    },
    "colab_type": "code",
    "id": "8Z1TNu5oC-f7",
    "outputId": "bb38234f-5592-439a-bed9-1e182a12663d"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/pandas/core/reshape/merge.py:522: UserWarning: merging between different levels can give an unintended result (2 levels on the left, 1 on the right)\n",
      "  warnings.warn(msg, UserWarning)\n"
     ]
    },
    {
     "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>(ALBUMIN, mean)</th>\n",
       "      <th>(ALBUMIN, std)</th>\n",
       "      <th>(ANIONGAP, mean)</th>\n",
       "      <th>(ANIONGAP, std)</th>\n",
       "      <th>(BANDS, mean)</th>\n",
       "      <th>(BANDS, std)</th>\n",
       "      <th>(BICARBONATE, mean)</th>\n",
       "      <th>(BICARBONATE, std)</th>\n",
       "      <th>(BILIRUBIN, mean)</th>\n",
       "      <th>(BILIRUBIN, std)</th>\n",
       "      <th>(BUN, mean)</th>\n",
       "      <th>(BUN, std)</th>\n",
       "      <th>(CHLORIDE, mean)</th>\n",
       "      <th>(CHLORIDE, std)</th>\n",
       "      <th>(CREATININE, mean)</th>\n",
       "      <th>(CREATININE, std)</th>\n",
       "      <th>(DBP, mean)</th>\n",
       "      <th>(DBP, std)</th>\n",
       "      <th>(EndoTrachFlag, mean)</th>\n",
       "      <th>(EndoTrachFlag, std)</th>\n",
       "      <th>(GCS, mean)</th>\n",
       "      <th>(GCS, std)</th>\n",
       "      <th>(GCSEyes, mean)</th>\n",
       "      <th>(GCSEyes, std)</th>\n",
       "      <th>(GCSMotor, mean)</th>\n",
       "      <th>(GCSMotor, std)</th>\n",
       "      <th>(GCSVerbal, mean)</th>\n",
       "      <th>(GCSVerbal, std)</th>\n",
       "      <th>(GLUCOSE, mean)</th>\n",
       "      <th>(GLUCOSE, std)</th>\n",
       "      <th>(HEMATOCRIT, mean)</th>\n",
       "      <th>(HEMATOCRIT, std)</th>\n",
       "      <th>(HEMOGLOBIN, mean)</th>\n",
       "      <th>(HEMOGLOBIN, std)</th>\n",
       "      <th>(HR, mean)</th>\n",
       "      <th>(HR, std)</th>\n",
       "      <th>(INR, mean)</th>\n",
       "      <th>(INR, std)</th>\n",
       "      <th>(LACTATE, mean)</th>\n",
       "      <th>(LACTATE, std)</th>\n",
       "      <th>(PLATELET, mean)</th>\n",
       "      <th>(PLATELET, std)</th>\n",
       "      <th>(POTASSIUM, mean)</th>\n",
       "      <th>(POTASSIUM, std)</th>\n",
       "      <th>(PT, mean)</th>\n",
       "      <th>(PT, std)</th>\n",
       "      <th>(PTT, mean)</th>\n",
       "      <th>(PTT, std)</th>\n",
       "      <th>(RSP, mean)</th>\n",
       "      <th>(RSP, std)</th>\n",
       "      <th>(SBP, mean)</th>\n",
       "      <th>(SBP, std)</th>\n",
       "      <th>(SODIUM, mean)</th>\n",
       "      <th>(SODIUM, std)</th>\n",
       "      <th>(SpO2, mean)</th>\n",
       "      <th>(SpO2, std)</th>\n",
       "      <th>(TMP, mean)</th>\n",
       "      <th>(TMP, std)</th>\n",
       "      <th>(WBC, mean)</th>\n",
       "      <th>(WBC, std)</th>\n",
       "      <th>(MAP, mean)</th>\n",
       "      <th>(MAP, std)</th>\n",
       "      <th>admission_age</th>\n",
       "      <th>gender</th>\n",
       "      <th>outcome</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>icustay_id</th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>200030</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>17.571429</td>\n",
       "      <td>4.157609</td>\n",
       "      <td>5.500000</td>\n",
       "      <td>3.535534</td>\n",
       "      <td>17.000000</td>\n",
       "      <td>1.632993</td>\n",
       "      <td>3.600000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>32.000000</td>\n",
       "      <td>7.238784</td>\n",
       "      <td>104.571429</td>\n",
       "      <td>6.948792</td>\n",
       "      <td>1.316667</td>\n",
       "      <td>0.453505</td>\n",
       "      <td>66.042373</td>\n",
       "      <td>7.609142</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>14.785714</td>\n",
       "      <td>0.578934</td>\n",
       "      <td>3.857143</td>\n",
       "      <td>0.363137</td>\n",
       "      <td>6.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>4.928571</td>\n",
       "      <td>0.267261</td>\n",
       "      <td>288.000000</td>\n",
       "      <td>160.031247</td>\n",
       "      <td>30.416667</td>\n",
       "      <td>2.394508</td>\n",
       "      <td>10.220000</td>\n",
       "      <td>0.875785</td>\n",
       "      <td>106.033898</td>\n",
       "      <td>14.534105</td>\n",
       "      <td>1.780000</td>\n",
       "      <td>0.164317</td>\n",
       "      <td>3.105882</td>\n",
       "      <td>1.732942</td>\n",
       "      <td>159.400000</td>\n",
       "      <td>17.728508</td>\n",
       "      <td>3.557143</td>\n",
       "      <td>0.571131</td>\n",
       "      <td>16.100000</td>\n",
       "      <td>0.667083</td>\n",
       "      <td>28.740000</td>\n",
       "      <td>1.594679</td>\n",
       "      <td>22.877193</td>\n",
       "      <td>4.799815</td>\n",
       "      <td>123.669492</td>\n",
       "      <td>14.526614</td>\n",
       "      <td>135.714286</td>\n",
       "      <td>4.785892</td>\n",
       "      <td>94.913793</td>\n",
       "      <td>2.842562</td>\n",
       "      <td>36.777782</td>\n",
       "      <td>0.787947</td>\n",
       "      <td>12.380000</td>\n",
       "      <td>1.907092</td>\n",
       "      <td>85.251412</td>\n",
       "      <td>9.394829</td>\n",
       "      <td>54</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200036</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23.333333</td>\n",
       "      <td>4.509250</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>49.666667</td>\n",
       "      <td>20.744477</td>\n",
       "      <td>113.333333</td>\n",
       "      <td>3.511885</td>\n",
       "      <td>1.566667</td>\n",
       "      <td>0.665833</td>\n",
       "      <td>75.081081</td>\n",
       "      <td>13.064750</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>11.333333</td>\n",
       "      <td>3.446562</td>\n",
       "      <td>3.333333</td>\n",
       "      <td>1.154701</td>\n",
       "      <td>5.000000</td>\n",
       "      <td>1.348400</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>1.348400</td>\n",
       "      <td>178.000000</td>\n",
       "      <td>151.792622</td>\n",
       "      <td>37.300000</td>\n",
       "      <td>5.367495</td>\n",
       "      <td>12.533333</td>\n",
       "      <td>1.975686</td>\n",
       "      <td>106.244898</td>\n",
       "      <td>14.706193</td>\n",
       "      <td>1.350000</td>\n",
       "      <td>0.070711</td>\n",
       "      <td>2.100000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>154.000000</td>\n",
       "      <td>25.534291</td>\n",
       "      <td>3.933333</td>\n",
       "      <td>0.838650</td>\n",
       "      <td>15.250000</td>\n",
       "      <td>0.636396</td>\n",
       "      <td>37.300000</td>\n",
       "      <td>9.758074</td>\n",
       "      <td>21.979167</td>\n",
       "      <td>3.900298</td>\n",
       "      <td>118.027027</td>\n",
       "      <td>17.147216</td>\n",
       "      <td>145.666667</td>\n",
       "      <td>8.082904</td>\n",
       "      <td>98.977778</td>\n",
       "      <td>1.453661</td>\n",
       "      <td>36.273506</td>\n",
       "      <td>0.205960</td>\n",
       "      <td>7.666667</td>\n",
       "      <td>2.494661</td>\n",
       "      <td>89.396396</td>\n",
       "      <td>13.384125</td>\n",
       "      <td>75</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200063</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13.250000</td>\n",
       "      <td>2.872281</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23.750000</td>\n",
       "      <td>2.986079</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>1.825742</td>\n",
       "      <td>106.750000</td>\n",
       "      <td>2.217356</td>\n",
       "      <td>1.275000</td>\n",
       "      <td>0.330404</td>\n",
       "      <td>52.716418</td>\n",
       "      <td>5.224668</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>NaN</td>\n",
       "      <td>115.333333</td>\n",
       "      <td>16.165808</td>\n",
       "      <td>27.320000</td>\n",
       "      <td>0.965401</td>\n",
       "      <td>9.525000</td>\n",
       "      <td>0.386221</td>\n",
       "      <td>91.984848</td>\n",
       "      <td>4.925577</td>\n",
       "      <td>1.166667</td>\n",
       "      <td>0.115470</td>\n",
       "      <td>2.283333</td>\n",
       "      <td>1.322750</td>\n",
       "      <td>171.750000</td>\n",
       "      <td>39.373214</td>\n",
       "      <td>4.225000</td>\n",
       "      <td>0.450000</td>\n",
       "      <td>13.733333</td>\n",
       "      <td>0.929157</td>\n",
       "      <td>31.833333</td>\n",
       "      <td>4.101626</td>\n",
       "      <td>16.891304</td>\n",
       "      <td>3.881394</td>\n",
       "      <td>111.119403</td>\n",
       "      <td>9.217116</td>\n",
       "      <td>139.250000</td>\n",
       "      <td>2.500000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>37.888889</td>\n",
       "      <td>0.505118</td>\n",
       "      <td>7.525000</td>\n",
       "      <td>1.887459</td>\n",
       "      <td>72.184080</td>\n",
       "      <td>5.798317</td>\n",
       "      <td>37</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200065</th>\n",
       "      <td>2.25</td>\n",
       "      <td>0.777817</td>\n",
       "      <td>24.500000</td>\n",
       "      <td>3.849439</td>\n",
       "      <td>23.333333</td>\n",
       "      <td>9.451631</td>\n",
       "      <td>17.166667</td>\n",
       "      <td>5.166911</td>\n",
       "      <td>1.450000</td>\n",
       "      <td>0.645497</td>\n",
       "      <td>28.333333</td>\n",
       "      <td>6.414945</td>\n",
       "      <td>102.583333</td>\n",
       "      <td>5.071459</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>0.612001</td>\n",
       "      <td>81.180328</td>\n",
       "      <td>10.788123</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>3.769231</td>\n",
       "      <td>1.877669</td>\n",
       "      <td>1.153846</td>\n",
       "      <td>0.375534</td>\n",
       "      <td>1.615385</td>\n",
       "      <td>1.502135</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>212.333333</td>\n",
       "      <td>68.063250</td>\n",
       "      <td>24.754545</td>\n",
       "      <td>2.055901</td>\n",
       "      <td>8.180000</td>\n",
       "      <td>0.614275</td>\n",
       "      <td>109.016393</td>\n",
       "      <td>14.606496</td>\n",
       "      <td>2.525000</td>\n",
       "      <td>1.019454</td>\n",
       "      <td>11.005000</td>\n",
       "      <td>2.611206</td>\n",
       "      <td>29.600000</td>\n",
       "      <td>23.716848</td>\n",
       "      <td>4.400000</td>\n",
       "      <td>0.718584</td>\n",
       "      <td>19.175000</td>\n",
       "      <td>3.918363</td>\n",
       "      <td>46.100000</td>\n",
       "      <td>20.122624</td>\n",
       "      <td>34.523810</td>\n",
       "      <td>5.679923</td>\n",
       "      <td>119.983607</td>\n",
       "      <td>22.596085</td>\n",
       "      <td>140.083333</td>\n",
       "      <td>2.020726</td>\n",
       "      <td>97.966667</td>\n",
       "      <td>2.603561</td>\n",
       "      <td>36.746918</td>\n",
       "      <td>1.668901</td>\n",
       "      <td>41.190000</td>\n",
       "      <td>6.842587</td>\n",
       "      <td>94.114754</td>\n",
       "      <td>14.270649</td>\n",
       "      <td>45</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200075</th>\n",
       "      <td>2.70</td>\n",
       "      <td>0.282843</td>\n",
       "      <td>15.500000</td>\n",
       "      <td>0.707107</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>16.333333</td>\n",
       "      <td>1.154701</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>1.272792</td>\n",
       "      <td>12.000000</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>108.000000</td>\n",
       "      <td>1.632993</td>\n",
       "      <td>0.733333</td>\n",
       "      <td>0.152753</td>\n",
       "      <td>56.000000</td>\n",
       "      <td>14.048158</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>NaN</td>\n",
       "      <td>83.666667</td>\n",
       "      <td>21.079216</td>\n",
       "      <td>31.166667</td>\n",
       "      <td>1.665333</td>\n",
       "      <td>11.000000</td>\n",
       "      <td>0.529150</td>\n",
       "      <td>80.180328</td>\n",
       "      <td>9.427810</td>\n",
       "      <td>1.275000</td>\n",
       "      <td>0.095743</td>\n",
       "      <td>1.680000</td>\n",
       "      <td>0.342053</td>\n",
       "      <td>124.666667</td>\n",
       "      <td>19.295941</td>\n",
       "      <td>3.725000</td>\n",
       "      <td>0.788987</td>\n",
       "      <td>14.575000</td>\n",
       "      <td>1.325079</td>\n",
       "      <td>66.033333</td>\n",
       "      <td>28.446769</td>\n",
       "      <td>21.438596</td>\n",
       "      <td>3.802103</td>\n",
       "      <td>114.639706</td>\n",
       "      <td>22.131534</td>\n",
       "      <td>136.250000</td>\n",
       "      <td>2.986079</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>36.459596</td>\n",
       "      <td>0.404520</td>\n",
       "      <td>23.550000</td>\n",
       "      <td>10.111627</td>\n",
       "      <td>75.546569</td>\n",
       "      <td>16.124049</td>\n",
       "      <td>83</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200087</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>10.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>6.666667</td>\n",
       "      <td>4.932883</td>\n",
       "      <td>23.333333</td>\n",
       "      <td>1.154701</td>\n",
       "      <td>0.700000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>33.666667</td>\n",
       "      <td>2.516611</td>\n",
       "      <td>109.333333</td>\n",
       "      <td>2.309401</td>\n",
       "      <td>0.633333</td>\n",
       "      <td>0.152753</td>\n",
       "      <td>60.828571</td>\n",
       "      <td>13.113724</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>NaN</td>\n",
       "      <td>175.333333</td>\n",
       "      <td>69.579691</td>\n",
       "      <td>30.380000</td>\n",
       "      <td>8.196768</td>\n",
       "      <td>10.433333</td>\n",
       "      <td>3.900427</td>\n",
       "      <td>102.886792</td>\n",
       "      <td>25.636258</td>\n",
       "      <td>1.400000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>1.316667</td>\n",
       "      <td>0.661564</td>\n",
       "      <td>197.666667</td>\n",
       "      <td>39.803685</td>\n",
       "      <td>3.866667</td>\n",
       "      <td>0.115470</td>\n",
       "      <td>15.800000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>116.133333</td>\n",
       "      <td>46.125626</td>\n",
       "      <td>23.377358</td>\n",
       "      <td>5.914378</td>\n",
       "      <td>105.600000</td>\n",
       "      <td>17.536852</td>\n",
       "      <td>138.666667</td>\n",
       "      <td>1.154701</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>36.569444</td>\n",
       "      <td>0.624565</td>\n",
       "      <td>12.900000</td>\n",
       "      <td>2.357965</td>\n",
       "      <td>75.752381</td>\n",
       "      <td>12.949208</td>\n",
       "      <td>64</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200102</th>\n",
       "      <td>3.15</td>\n",
       "      <td>0.353553</td>\n",
       "      <td>19.714286</td>\n",
       "      <td>3.450328</td>\n",
       "      <td>3.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>23.142857</td>\n",
       "      <td>1.864454</td>\n",
       "      <td>40.114286</td>\n",
       "      <td>2.735829</td>\n",
       "      <td>82.750000</td>\n",
       "      <td>6.341473</td>\n",
       "      <td>101.000000</td>\n",
       "      <td>1.154701</td>\n",
       "      <td>2.450000</td>\n",
       "      <td>0.333809</td>\n",
       "      <td>51.522727</td>\n",
       "      <td>8.790918</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.347826</td>\n",
       "      <td>2.854163</td>\n",
       "      <td>2.086957</td>\n",
       "      <td>1.083473</td>\n",
       "      <td>4.260870</td>\n",
       "      <td>1.959067</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>177.142857</td>\n",
       "      <td>54.241523</td>\n",
       "      <td>29.675000</td>\n",
       "      <td>1.401785</td>\n",
       "      <td>10.312500</td>\n",
       "      <td>0.491172</td>\n",
       "      <td>78.151515</td>\n",
       "      <td>5.778588</td>\n",
       "      <td>1.837500</td>\n",
       "      <td>0.184681</td>\n",
       "      <td>3.880000</td>\n",
       "      <td>1.411382</td>\n",
       "      <td>82.625000</td>\n",
       "      <td>32.137817</td>\n",
       "      <td>4.285714</td>\n",
       "      <td>0.353217</td>\n",
       "      <td>16.587500</td>\n",
       "      <td>0.821910</td>\n",
       "      <td>39.837500</td>\n",
       "      <td>4.098759</td>\n",
       "      <td>19.735294</td>\n",
       "      <td>1.997145</td>\n",
       "      <td>115.204545</td>\n",
       "      <td>17.056011</td>\n",
       "      <td>139.428571</td>\n",
       "      <td>0.975900</td>\n",
       "      <td>95.666667</td>\n",
       "      <td>1.379116</td>\n",
       "      <td>36.027778</td>\n",
       "      <td>0.507942</td>\n",
       "      <td>16.350000</td>\n",
       "      <td>2.949092</td>\n",
       "      <td>72.750000</td>\n",
       "      <td>11.233392</td>\n",
       "      <td>46</td>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200140</th>\n",
       "      <td>2.60</td>\n",
       "      <td>NaN</td>\n",
       "      <td>13.600000</td>\n",
       "      <td>2.880972</td>\n",
       "      <td>4.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>33.200000</td>\n",
       "      <td>4.024922</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>65.800000</td>\n",
       "      <td>14.601370</td>\n",
       "      <td>105.200000</td>\n",
       "      <td>3.898718</td>\n",
       "      <td>1.700000</td>\n",
       "      <td>0.275681</td>\n",
       "      <td>65.345070</td>\n",
       "      <td>9.876396</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>7.791667</td>\n",
       "      <td>1.178767</td>\n",
       "      <td>2.666667</td>\n",
       "      <td>0.963087</td>\n",
       "      <td>4.125000</td>\n",
       "      <td>0.337832</td>\n",
       "      <td>1.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>118.500000</td>\n",
       "      <td>33.756481</td>\n",
       "      <td>48.450000</td>\n",
       "      <td>7.696536</td>\n",
       "      <td>15.100000</td>\n",
       "      <td>2.024846</td>\n",
       "      <td>78.077465</td>\n",
       "      <td>12.257332</td>\n",
       "      <td>1.220000</td>\n",
       "      <td>0.277489</td>\n",
       "      <td>2.225000</td>\n",
       "      <td>0.556028</td>\n",
       "      <td>183.000000</td>\n",
       "      <td>29.743347</td>\n",
       "      <td>3.866667</td>\n",
       "      <td>0.233809</td>\n",
       "      <td>14.280000</td>\n",
       "      <td>2.758985</td>\n",
       "      <td>61.580000</td>\n",
       "      <td>46.123335</td>\n",
       "      <td>14.972028</td>\n",
       "      <td>0.454927</td>\n",
       "      <td>101.623239</td>\n",
       "      <td>19.921939</td>\n",
       "      <td>147.333333</td>\n",
       "      <td>4.366539</td>\n",
       "      <td>97.125984</td>\n",
       "      <td>2.569640</td>\n",
       "      <td>36.356483</td>\n",
       "      <td>0.692044</td>\n",
       "      <td>9.825000</td>\n",
       "      <td>5.463439</td>\n",
       "      <td>77.437793</td>\n",
       "      <td>12.837280</td>\n",
       "      <td>68</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200215</th>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>16.000000</td>\n",
       "      <td>2.581989</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>21.750000</td>\n",
       "      <td>2.061553</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>42.750000</td>\n",
       "      <td>1.258306</td>\n",
       "      <td>104.250000</td>\n",
       "      <td>0.500000</td>\n",
       "      <td>1.300000</td>\n",
       "      <td>0.141421</td>\n",
       "      <td>58.645455</td>\n",
       "      <td>11.630369</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>NaN</td>\n",
       "      <td>103.750000</td>\n",
       "      <td>12.120919</td>\n",
       "      <td>34.100000</td>\n",
       "      <td>2.605763</td>\n",
       "      <td>11.300000</td>\n",
       "      <td>0.793725</td>\n",
       "      <td>98.076923</td>\n",
       "      <td>8.526421</td>\n",
       "      <td>1.900000</td>\n",
       "      <td>0.424264</td>\n",
       "      <td>1.833333</td>\n",
       "      <td>0.378594</td>\n",
       "      <td>413.666667</td>\n",
       "      <td>34.486712</td>\n",
       "      <td>4.875000</td>\n",
       "      <td>0.694622</td>\n",
       "      <td>20.650000</td>\n",
       "      <td>3.464823</td>\n",
       "      <td>36.300000</td>\n",
       "      <td>1.414214</td>\n",
       "      <td>17.884615</td>\n",
       "      <td>4.945403</td>\n",
       "      <td>106.254545</td>\n",
       "      <td>12.752586</td>\n",
       "      <td>137.000000</td>\n",
       "      <td>2.000000</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>36.419192</td>\n",
       "      <td>0.603162</td>\n",
       "      <td>20.300000</td>\n",
       "      <td>2.253886</td>\n",
       "      <td>74.515152</td>\n",
       "      <td>10.675742</td>\n",
       "      <td>84</td>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>200231</th>\n",
       "      <td>1.50</td>\n",
       "      <td>NaN</td>\n",
       "      <td>15.000000</td>\n",
       "      <td>2.160247</td>\n",
       "      <td>26.000000</td>\n",
       "      <td>8.755950</td>\n",
       "      <td>19.250000</td>\n",
       "      <td>0.957427</td>\n",
       "      <td>4.366667</td>\n",
       "      <td>0.351188</td>\n",
       "      <td>30.000000</td>\n",
       "      <td>3.559026</td>\n",
       "      <td>101.750000</td>\n",
       "      <td>3.500000</td>\n",
       "      <td>2.300000</td>\n",
       "      <td>0.216025</td>\n",
       "      <td>46.185185</td>\n",
       "      <td>8.413991</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>NaN</td>\n",
       "      <td>86.250000</td>\n",
       "      <td>15.734781</td>\n",
       "      <td>29.325000</td>\n",
       "      <td>1.342572</td>\n",
       "      <td>8.975000</td>\n",
       "      <td>0.403113</td>\n",
       "      <td>78.763636</td>\n",
       "      <td>6.155286</td>\n",
       "      <td>3.033333</td>\n",
       "      <td>0.404145</td>\n",
       "      <td>5.083333</td>\n",
       "      <td>1.099848</td>\n",
       "      <td>60.250000</td>\n",
       "      <td>5.188127</td>\n",
       "      <td>4.300000</td>\n",
       "      <td>0.182574</td>\n",
       "      <td>31.166667</td>\n",
       "      <td>3.982880</td>\n",
       "      <td>54.800000</td>\n",
       "      <td>14.142136</td>\n",
       "      <td>23.400000</td>\n",
       "      <td>3.634200</td>\n",
       "      <td>101.833333</td>\n",
       "      <td>14.287810</td>\n",
       "      <td>131.750000</td>\n",
       "      <td>2.217356</td>\n",
       "      <td>NaN</td>\n",
       "      <td>NaN</td>\n",
       "      <td>36.585859</td>\n",
       "      <td>0.199325</td>\n",
       "      <td>2.550000</td>\n",
       "      <td>1.181807</td>\n",
       "      <td>64.734568</td>\n",
       "      <td>9.519126</td>\n",
       "      <td>73</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "            (ALBUMIN, mean)  (ALBUMIN, std)  ...  gender  outcome\n",
       "icustay_id                                   ...                 \n",
       "200030                  NaN             NaN  ...       1        0\n",
       "200036                  NaN             NaN  ...       1        0\n",
       "200063                  NaN             NaN  ...       1        0\n",
       "200065                 2.25        0.777817  ...       0        0\n",
       "200075                 2.70        0.282843  ...       0        0\n",
       "200087                  NaN             NaN  ...       1        0\n",
       "200102                 3.15        0.353553  ...       1        1\n",
       "200140                 2.60             NaN  ...       1        0\n",
       "200215                  NaN             NaN  ...       1        0\n",
       "200231                 1.50             NaN  ...       0        1\n",
       "\n",
       "[10 rows x 65 columns]"
      ]
     },
     "execution_count": 128,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ds = ts_df.drop('delta', axis = 1).groupby('icustay_id').agg(['mean','std']) \n",
    "ds = ds.merge(cohort[['icustay_id', 'admission_age', 'gender','hospital_expire_flag']], on = 'icustay_id').rename({'hospital_expire_flag': 'outcome'}, axis = 1).set_index('icustay_id')\n",
    "\n",
    "ds.head(10)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Zb6-DlpLqTD8"
   },
   "source": [
    "How to expand the features space beside mean and stadard deviaton? "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qbnsTJS6XbN3"
   },
   "source": [
    "**G: Modeling**\n",
    "\n",
    "Once the dataset is ready we can attempt a simple classification. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "xSXTevkcHpXc"
   },
   "outputs": [],
   "source": [
    "# useful libraries\n",
    "\n",
    "from sklearn.model_selection import train_test_split, RandomizedSearchCV, GridSearchCV\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "from sklearn.ensemble import RandomForestClassifier, GradientBoostingClassifier\n",
    "from sklearn.impute import SimpleImputer\n",
    "from sklearn.preprocessing import StandardScaler\n",
    "from sklearn.pipeline import Pipeline\n",
    "import sklearn.metrics as mtr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "2qpAcw2AKh4N"
   },
   "outputs": [],
   "source": [
    "# useful functions\n",
    "\n",
    "\n",
    "# Plots roc curve\n",
    "def plot_roc(y_true, y_proba, pos_class = 1):\n",
    "  fpr, tpr, _ = mtr.roc_curve(y_true, y_proba[:,pos_class])\n",
    "  auc = mtr.auc(fpr, tpr)\n",
    "  \n",
    "  fig, ax = plt.subplots()\n",
    "  ax.plot(fpr, tpr)\n",
    "  ax.plot([0,1], [0, 1], color = 'k', linestyle = '--', alpha = .3)\n",
    "  ax.text(.5, .5, 'AUC: {:.2f}'.format(auc), fontweight = 'bold', ha = 'center')\n",
    "  \n",
    "  ax.grid(alpha = .3)\n",
    "  ax.set_xlim([0,1])\n",
    "  ax.set_ylim([0,1])\n",
    "  ax.set_xlabel('FPR')\n",
    "  ax.set_ylabel('TPR')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "L4_IMitbD8Kp"
   },
   "outputs": [],
   "source": [
    "# Split train and test set in  75%/25% \n",
    "\n",
    "X, y = ds.drop('outcome', axis = 1), ds.outcome\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, random_state = 0, test_size = .25, stratify = y)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 317
    },
    "colab_type": "code",
    "id": "745BcylCIe3e",
    "outputId": "082e7e14-802f-4b46-b671-799cdaf964ff"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/sklearn/linear_model/logistic.py:432: FutureWarning: Default solver will be changed to 'lbfgs' in 0.22. Specify a solver to silence this warning.\n",
      "  FutureWarning)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Logistic Regression \n",
    "\n",
    "clf_lr = Pipeline([\n",
    "    ('si', SimpleImputer()),\n",
    "    ('sd', StandardScaler()),\n",
    "    ('lr', LogisticRegression(random_state = 0))\n",
    "])\n",
    "\n",
    "clf_lr.fit(X_train, y_train)\n",
    "y_prob = clf_lr.predict_proba(X_test)\n",
    "plot_roc(y_test, y_prob)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "sSyB0ZzuuBdr",
    "outputId": "beac5d82-fcd6-4b04-bc9b-a825cc4b0b48"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.74      0.85      0.79       571\n",
      "           1       0.65      0.48      0.56       331\n",
      "\n",
      "    accuracy                           0.72       902\n",
      "   macro avg       0.70      0.67      0.67       902\n",
      "weighted avg       0.71      0.72      0.70       902\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = clf_lr.predict(X_test)\n",
    "print(mtr.classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 718
    },
    "colab_type": "code",
    "id": "rH-_5RiH12Dl",
    "outputId": "a7080827-009d-4d37-e25f-7b84fb41cecd"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7f1ecdd26c50>]"
      ]
     },
     "execution_count": 181,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "odds = np.exp(clf_lr.steps[2][1].coef_)\n",
    "\n",
    "feat_names = [val[0] + val[1] for val in ds.drop('outcome', axis = 1).columns]\n",
    "feat_df = pd.DataFrame({'feature': feat_names, 'odds': odds[0]}).sort_values(by = 'odds', ascending = False)\n",
    "\n",
    "fig, ax = plt.subplots(figsize = (20, 10))\n",
    "feat_df.plot.bar(x = 'feature', ax = ax)\n",
    "ax.plot(ax.get_xlim(), [1,1], color = 'k', linestyle = '--', linewidth = 2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "n95YoUxOwFrN"
   },
   "source": [
    "* Does this model overfit?\n",
    "* Are features all indipendent?\n",
    "* Would feature selection improve perfromance?\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 283
    },
    "colab_type": "code",
    "id": "JsYQ3M3rMmoL",
    "outputId": "2443c368-c2fb-49d8-bfcf-1bf73f09dda6"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Random Forest\n",
    "\n",
    "clf_rf = Pipeline([\n",
    "    ('si', SimpleImputer()),\n",
    "    ('rf', RandomForestClassifier(random_state = 0))\n",
    "])\n",
    "\n",
    "random_param = {\n",
    "    'rf__n_estimators': np.arange(100, 1000, 100),\n",
    "    'rf__max_depth': [10, 20, 30, 40, 50, 60, 70, 80, 90, 100, None],\n",
    "    'rf__max_features': ['auto', 'sqrt'],\n",
    "    'rf__min_samples_leaf': [1, 2, 4],\n",
    "    'rf__min_samples_split': [2, 5, 10]\n",
    "}\n",
    "\n",
    "rs = RandomizedSearchCV(clf_rf, random_param, random_state = 0, cv = 10, \n",
    "                        n_iter = 10, n_jobs = -1)\n",
    "rs.fit(X_train, y_train)\n",
    "y_prob = rs.best_estimator_.predict_proba(X_test)\n",
    "plot_roc(y_test, y_prob)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 170
    },
    "colab_type": "code",
    "id": "kqI6pB5XtMj1",
    "outputId": "2fbdb339-0139-460e-c089-3c96f37eb1aa"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.73      0.90      0.80       571\n",
      "           1       0.71      0.41      0.52       331\n",
      "\n",
      "    accuracy                           0.72       902\n",
      "   macro avg       0.72      0.66      0.66       902\n",
      "weighted avg       0.72      0.72      0.70       902\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = rs.best_estimator_.predict(X_test)\n",
    "print(mtr.classification_report(y_test, y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 718
    },
    "colab_type": "code",
    "id": "KTsfhjKh5DZu",
    "outputId": "22f89ccf-89a0-4263-fd02-961fe79a2a56"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f1ed77b85f8>"
      ]
     },
     "execution_count": 175,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "rank = rs.best_estimator_.steps[1][1].feature_importances_\n",
    "rf_rank = pd.DataFrame({'feature': feat_names, 'rank': rank}).sort_values(by = 'rank', ascending = False)\n",
    "\n",
    "\n",
    "fig, ax = plt.subplots(figsize = (20, 10))\n",
    "rf_rank.plot.bar(x = 'feature', ax = ax)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "r8PRh3jJ-lEf"
   },
   "source": [
    "* Is random forest overfitting?\n",
    "* How does it compare to gradient boosting?"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "nsI4nG8oCygi"
   },
   "source": [
    "[1] Early Prediction of Sepsis from Clinical Data -- the PhysioNet Computing in Cardiology Challenge 2019, M Reyna et al \n",
    "\n",
    "[2] Vincent, Jean-Louis, et al. \"Sepsis definitions: time for change.\" Lancet (London, England) 381.9868 (2013): 774.\n",
    "\n",
    "[3] Singer, Mervyn, et al. \"The third international consensus definitions for sepsis and septic shock (Sepsis-3).\" Jama 315.8 (2016): 801-810.\n",
    "\n",
    "[4] Pirracchio, Romain, et al. \"Mortality prediction in intensive care units with the Super ICU Learner Algorithm (SICULA): a population-based study.\" The Lancet Respiratory Medicine 3.1 (2015): 42-52"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "mortality_in_septic_patients.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}