First solution.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xwi9mzFPPDv7"
      },
      "source": [
        "# Identifying Entities in Healthcare Data"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3wwzoMzrUDtr"
      },
      "source": [
        "##Workspace set up: Import and Install useful packages."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "T4hxafMCE1nn",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [],
      "source": [
        "# !pip install pycrf\n",
        "# !pip install sklearn-crfsuite\n",
        "\n",
        "import pandas as pd\n",
        "import warnings\n",
        "warnings.filterwarnings('ignore')\n",
        "\n",
        "import re\n",
        "import spacy\n",
        "\n",
        "import sklearn_crfsuite\n",
        "from sklearn_crfsuite import metrics\n",
        "\n",
        "model = spacy.load(\"en_core_web_trf\")"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bE9Hwb3AUHwG"
      },
      "source": [
        "##Data Preprocessing\n",
        "\n",
        "The dataset provided is in the form of one word per line. Let's understand the format of data below:\n",
        "- Suppose there are *x* words in a sentence, then there will be *x* continuous lines with one word in each line. \n",
        "- Further, the two sentences are separated by empty lines. The labels for the data follow the same format.\n",
        "\n",
        "**We need to pre-process the data to recover the complete sentences and their labels.**\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nBU_BNUuK8tm"
      },
      "source": [
        "### Construct the proper sentences from individual words and print the 5 sentences."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "b4OjXEhjhkCW"
      },
      "source": [
        "![image.png](data:image/png;base64,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)\n",
        "\n",
        "![image.png](data:image/png;base64,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)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "bz6czVx9R6WJ",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [],
      "source": [
        "# Function to read sentence and label and return as a pandas series\n",
        "def read_data(sent_file, label_file):\n",
        "    sent, label = [], []\n",
        "    with open(sent_file, 'r') as sents, open(label_file, 'r') as labels:\n",
        "        sent_temp, labels_temp = '',''\n",
        "        for sent_line, label_line in zip(sents, labels):\n",
        "            sent_line, label_line = sent_line.strip(), label_line.strip()\n",
        "            if sent_line!='' and label_line!='' and re.match(r'^[A-Za-z0-9].*$', sent_line):\n",
        "                sent_temp += f'{sent_line} '\n",
        "                labels_temp += f'{label_line} '\n",
        "            elif sent_line=='' and label_line=='':\n",
        "                sent.append(sent_temp)\n",
        "                label.append(labels_temp)\n",
        "                sent_temp, labels_temp = '',''\n",
        "    return sent,label"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "id": "TE31q8yxVdU-",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [],
      "source": [
        "train_sent,train_label=read_data('train_sent', 'train_label')\n",
        "test_sent,test_label=read_data('test_sent', 'test_label')"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "### Pritning first five sentences with label from training data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "k4gdPBHwLdNX",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "All:O live:O births:O or:O 23:O weeks:O at:O the:O University:O of:O Vermont:O in:O 1995:O n:O 2395:O were:O retrospectively:O analyzed:O for:O delivery:O route:O indication:O for:O cesarean:O gestational:O age:O parity:O and:O practice:O group:O to:O reflect:O risk:O status:O : \n",
            "\n",
            "The:O total:O cesarean:O rate:O was:O 14.4:O 344:O of:O 2395:O and:O the:O primary:O rate:O was:O 11.4:O 244:O of:O 2144:O : \n",
            "\n",
            "Abnormal:O presentation:O was:O the:O most:O common:O indication:O 25.6:O 88:O of:O 344:O : \n",
            "\n",
            "The:O corrected:O cesarean:O rate:O maternal-fetal:O medicine:O and:O transported:O patients:O excluded:O was:O 12.4:O 273:O of:O 2194:O and:O the:O corrected:O primary:O rate:O was:O 9.6:O 190:O of:O 1975:O : \n",
            "\n",
            "Arrest:O of:O dilation:O was:O the:O most:O common:O indication:O in:O both:O corrected:O subgroups:O 23.4:O and:O 24.6:O respectively:O : \n",
            "\n"
          ]
        }
      ],
      "source": [
        "for i in range(5):\n",
        "  for word,label in zip(train_sent[i].split(' '),train_label[i].split(' ')):\n",
        "    print(f'{word}:{label}',sep=' ',end=' ')\n",
        "  print('\\n')"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "### Pritning first five sentences with label from test data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "k4gdPBHwLdNX",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Furthermore:O when:O all:O deliveries:O were:O analyzed:O regardless:O of:O risk:O status:O but:O limited:O to:O gestational:O age:O or:O 36:O weeks:O the:O rates:O did:O not:O change:O 12.6:O 280:O of:O 2214:O primary:O 9.2:O 183:O of:O 1994:O : \n",
            "\n",
            "As:O the:O ambient:O temperature:O increases:O there:O is:O an:O increase:O in:O insensible:O fluid:O loss:O and:O the:O potential:O for:O dehydration:O : \n",
            "\n",
            "The:O daily:O high:O temperature:O ranged:O from:O 71:O to:O 104:O degrees:O F:O and:O AFI:O values:O ranged:O from:O 1.7:O to:O 24.7:O cm:O during:O the:O study:O period:O : \n",
            "\n",
            "There:O was:O a:O significant:O correlation:O between:O the:O 2-:O 3-:O and:O 4-day:O mean:O temperature:O and:O AFI:O with:O the:O 4-day:O mean:O being:O the:O most:O significant:O r:O 0.31:O p:O 60:O 0.001:O : \n",
            "\n",
            "Fluctuations:O in:O ambient:O temperature:O are:O inversely:O correlated:O to:O changes:O in:O AFI:O : \n",
            "\n"
          ]
        }
      ],
      "source": [
        "for i in range(5):\n",
        "  for word,label in zip(test_sent[i].split(' '),test_label[i].split(' ')):\n",
        "    print(f'{word}:{label}',sep=' ',end=' ')\n",
        "  print('\\n')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "urCxIOBTMBwG"
      },
      "source": [
        "### Count the number of sentences in the processed train and test dataset "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "zwa18H_IMFXV",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Count of sentences in Train: 2599\n",
            "Count of sentences in Test: 1056\n"
          ]
        }
      ],
      "source": [
        "print('Count of sentences in Train:',len(train_sent))\n",
        "print('Count of sentences in Test:',len(test_sent))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Jjg2DK7hLcO_"
      },
      "source": [
        "### Count the number of lines of labels in the processed train and test dataset."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "unXoD2bWLPZD",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Count of label in Train: 2599\n",
            "Count of label in Test: 1056\n"
          ]
        }
      ],
      "source": [
        "print('Count of label in Train:',len(train_label))\n",
        "print('Count of label in Test:',len(test_label))"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "### Preprocess data for Concept Identification "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [],
      "source": [
        "# function to return dataframe with word, label, lemma, pos, tag, dep, shape, is_alpha, is_stop\n",
        "def get_preprocessed_data(train_sent, train_label):\n",
        "    train_data = pd.DataFrame(columns=['word', 'label', 'lemma', 'pos', 'tag', 'dep', 'shape', 'is_alpha', 'is_stop'])\n",
        "    for sent, label in zip(train_sent, train_label):\n",
        "        sent, label = sent.strip(), label.strip()\n",
        "        doc = model(sent)\n",
        "        for token, label in zip(doc, label.split()):\n",
        "            train_data = train_data.append({'sent':sent,'word':token.text,'position':token.i,'label':label,'lemma':token.lemma_,'pos':token.pos_,'tag':token.tag_,'dep':token.dep_,'shape':token.shape_,'is_alpha':token.is_alpha,'is_stop':token.is_stop}, ignore_index=True)\n",
        "    return train_data"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [],
      "source": [
        "train_data = get_preprocessed_data(train_sent, train_label)\n",
        "test_data = get_preprocessed_data(test_sent, test_label)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 10,
      "metadata": {
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "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>word</th>\n",
              "      <th>label</th>\n",
              "      <th>lemma</th>\n",
              "      <th>pos</th>\n",
              "      <th>tag</th>\n",
              "      <th>dep</th>\n",
              "      <th>shape</th>\n",
              "      <th>is_alpha</th>\n",
              "      <th>is_stop</th>\n",
              "      <th>sent</th>\n",
              "      <th>position</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>All</td>\n",
              "      <td>O</td>\n",
              "      <td>all</td>\n",
              "      <td>DET</td>\n",
              "      <td>DT</td>\n",
              "      <td>det</td>\n",
              "      <td>Xxx</td>\n",
              "      <td>True</td>\n",
              "      <td>True</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>live</td>\n",
              "      <td>O</td>\n",
              "      <td>live</td>\n",
              "      <td>ADJ</td>\n",
              "      <td>JJ</td>\n",
              "      <td>amod</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>births</td>\n",
              "      <td>O</td>\n",
              "      <td>birth</td>\n",
              "      <td>NOUN</td>\n",
              "      <td>NNS</td>\n",
              "      <td>nsubjpass</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>2.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>or</td>\n",
              "      <td>O</td>\n",
              "      <td>or</td>\n",
              "      <td>CCONJ</td>\n",
              "      <td>CC</td>\n",
              "      <td>cc</td>\n",
              "      <td>xx</td>\n",
              "      <td>True</td>\n",
              "      <td>True</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>3.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>23</td>\n",
              "      <td>O</td>\n",
              "      <td>23</td>\n",
              "      <td>NUM</td>\n",
              "      <td>CD</td>\n",
              "      <td>nummod</td>\n",
              "      <td>dd</td>\n",
              "      <td>False</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>4.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "     word label  lemma    pos  tag        dep shape is_alpha is_stop  \\\n",
              "0     All     O    all    DET   DT        det   Xxx     True    True   \n",
              "1    live     O   live    ADJ   JJ       amod  xxxx     True   False   \n",
              "2  births     O  birth   NOUN  NNS  nsubjpass  xxxx     True   False   \n",
              "3      or     O     or  CCONJ   CC         cc    xx     True    True   \n",
              "4      23     O     23    NUM   CD     nummod    dd    False   False   \n",
              "\n",
              "                                                sent  position  \n",
              "0  All live births or 23 weeks at the University ...       0.0  \n",
              "1  All live births or 23 weeks at the University ...       1.0  \n",
              "2  All live births or 23 weeks at the University ...       2.0  \n",
              "3  All live births or 23 weeks at the University ...       3.0  \n",
              "4  All live births or 23 weeks at the University ...       4.0  "
            ]
          },
          "execution_count": 10,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "train_data.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 11,
      "metadata": {
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "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>word</th>\n",
              "      <th>label</th>\n",
              "      <th>lemma</th>\n",
              "      <th>pos</th>\n",
              "      <th>tag</th>\n",
              "      <th>dep</th>\n",
              "      <th>shape</th>\n",
              "      <th>is_alpha</th>\n",
              "      <th>is_stop</th>\n",
              "      <th>sent</th>\n",
              "      <th>position</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>Furthermore</td>\n",
              "      <td>O</td>\n",
              "      <td>furthermore</td>\n",
              "      <td>ADV</td>\n",
              "      <td>RB</td>\n",
              "      <td>advmod</td>\n",
              "      <td>Xxxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>Furthermore when all deliveries were analyzed ...</td>\n",
              "      <td>0.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>when</td>\n",
              "      <td>O</td>\n",
              "      <td>when</td>\n",
              "      <td>SCONJ</td>\n",
              "      <td>WRB</td>\n",
              "      <td>advmod</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>True</td>\n",
              "      <td>Furthermore when all deliveries were analyzed ...</td>\n",
              "      <td>1.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>all</td>\n",
              "      <td>O</td>\n",
              "      <td>all</td>\n",
              "      <td>DET</td>\n",
              "      <td>DT</td>\n",
              "      <td>det</td>\n",
              "      <td>xxx</td>\n",
              "      <td>True</td>\n",
              "      <td>True</td>\n",
              "      <td>Furthermore when all deliveries were analyzed ...</td>\n",
              "      <td>2.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>deliveries</td>\n",
              "      <td>O</td>\n",
              "      <td>delivery</td>\n",
              "      <td>NOUN</td>\n",
              "      <td>NNS</td>\n",
              "      <td>nsubjpass</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>Furthermore when all deliveries were analyzed ...</td>\n",
              "      <td>3.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>were</td>\n",
              "      <td>O</td>\n",
              "      <td>be</td>\n",
              "      <td>AUX</td>\n",
              "      <td>VBD</td>\n",
              "      <td>auxpass</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>True</td>\n",
              "      <td>Furthermore when all deliveries were analyzed ...</td>\n",
              "      <td>4.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "          word label        lemma    pos  tag        dep  shape is_alpha  \\\n",
              "0  Furthermore     O  furthermore    ADV   RB     advmod  Xxxxx     True   \n",
              "1         when     O         when  SCONJ  WRB     advmod   xxxx     True   \n",
              "2          all     O          all    DET   DT        det    xxx     True   \n",
              "3   deliveries     O     delivery   NOUN  NNS  nsubjpass   xxxx     True   \n",
              "4         were     O           be    AUX  VBD    auxpass   xxxx     True   \n",
              "\n",
              "  is_stop                                               sent  position  \n",
              "0   False  Furthermore when all deliveries were analyzed ...       0.0  \n",
              "1    True  Furthermore when all deliveries were analyzed ...       1.0  \n",
              "2    True  Furthermore when all deliveries were analyzed ...       2.0  \n",
              "3   False  Furthermore when all deliveries were analyzed ...       3.0  \n",
              "4    True  Furthermore when all deliveries were analyzed ...       4.0  "
            ]
          },
          "execution_count": 11,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "test_data.head()"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "r9Hy2tECbd9a"
      },
      "source": [
        "## Concept Identification\n",
        "\n",
        "We will first explore what are the various concepts present in the dataset. For this, we will use PoS Tagging. \n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [],
      "source": [
        "# Combining test and train for concept identification\n",
        "complete_data = pd.concat([train_data, test_data], ignore_index=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3YGpfOK7Mk3X"
      },
      "source": [
        "### Extract those tokens which have NOUN or PROPN as their PoS tag and find their frequency"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 13,
      "metadata": {
        "id": "7w8nJSPlZqjw",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "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>word</th>\n",
              "      <th>label</th>\n",
              "      <th>lemma</th>\n",
              "      <th>pos</th>\n",
              "      <th>tag</th>\n",
              "      <th>dep</th>\n",
              "      <th>shape</th>\n",
              "      <th>is_alpha</th>\n",
              "      <th>is_stop</th>\n",
              "      <th>sent</th>\n",
              "      <th>position</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>births</td>\n",
              "      <td>O</td>\n",
              "      <td>birth</td>\n",
              "      <td>NOUN</td>\n",
              "      <td>NNS</td>\n",
              "      <td>nsubjpass</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>2.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>weeks</td>\n",
              "      <td>O</td>\n",
              "      <td>week</td>\n",
              "      <td>NOUN</td>\n",
              "      <td>NNS</td>\n",
              "      <td>conj</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>5.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>University</td>\n",
              "      <td>O</td>\n",
              "      <td>University</td>\n",
              "      <td>PROPN</td>\n",
              "      <td>NNP</td>\n",
              "      <td>pobj</td>\n",
              "      <td>Xxxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>8.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>Vermont</td>\n",
              "      <td>O</td>\n",
              "      <td>Vermont</td>\n",
              "      <td>PROPN</td>\n",
              "      <td>NNP</td>\n",
              "      <td>pobj</td>\n",
              "      <td>Xxxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>10.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>delivery</td>\n",
              "      <td>O</td>\n",
              "      <td>delivery</td>\n",
              "      <td>NOUN</td>\n",
              "      <td>NN</td>\n",
              "      <td>compound</td>\n",
              "      <td>xxxx</td>\n",
              "      <td>True</td>\n",
              "      <td>False</td>\n",
              "      <td>All live births or 23 weeks at the University ...</td>\n",
              "      <td>19.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "          word label       lemma    pos  tag        dep  shape is_alpha  \\\n",
              "2       births     O       birth   NOUN  NNS  nsubjpass   xxxx     True   \n",
              "5        weeks     O        week   NOUN  NNS       conj   xxxx     True   \n",
              "8   University     O  University  PROPN  NNP       pobj  Xxxxx     True   \n",
              "10     Vermont     O     Vermont  PROPN  NNP       pobj  Xxxxx     True   \n",
              "19    delivery     O    delivery   NOUN   NN   compound   xxxx     True   \n",
              "\n",
              "   is_stop                                               sent  position  \n",
              "2    False  All live births or 23 weeks at the University ...       2.0  \n",
              "5    False  All live births or 23 weeks at the University ...       5.0  \n",
              "8    False  All live births or 23 weeks at the University ...       8.0  \n",
              "10   False  All live births or 23 weeks at the University ...      10.0  \n",
              "19   False  All live births or 23 weeks at the University ...      19.0  "
            ]
          },
          "execution_count": 13,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# Extracting rows with pos tag as NOUN, PROPN\n",
        "concept_data = complete_data[complete_data['pos'].isin(['NOUN', 'PROPN'])]\n",
        "concept_data.head()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-SQG8wRxMo6A"
      },
      "source": [
        "### Print the top 25 most common tokens with NOUN or PROPN PoS tags"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "id": "DhDY4jjoawTa",
        "notebookRunGroups": {
          "groupValue": "1"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "patients        465\n",
              "treatment       272\n",
              "cancer          165\n",
              "therapy         154\n",
              "study           142\n",
              "cell            118\n",
              "disease         116\n",
              "lung             94\n",
              "group            85\n",
              "effects          84\n",
              "gene             79\n",
              "chemotherapy     78\n",
              "use              73\n",
              "women            73\n",
              "results          71\n",
              "cases            71\n",
              "patient          70\n",
              "-                69\n",
              "surgery          68\n",
              "risk             66\n",
              "rate             66\n",
              "response         64\n",
              "children         62\n",
              "effect           62\n",
              "analysis         60\n",
              "Name: word, dtype: int64"
            ]
          },
          "execution_count": 14,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# Printing the top 25 concepts with their count\n",
        "concept_data['word'].value_counts().head(25)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bnfAlKgA5fd4"
      },
      "source": [
        "## Defining features for CRF\n",
        "\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "aS6P1Gog3JMf",
        "notebookRunGroups": {
          "groupValue": "2"
        }
      },
      "outputs": [],
      "source": [
        "# function to take a take a row of dataframe and return a list of features\n",
        "def getFeaturesForOneWord(row,data):\n",
        "  features = [\n",
        "      f'word:{row[\"word\"]}',\n",
        "      f'wordword.lower={row[\"word\"].lower()}',\n",
        "      f'word[-3:]={row[\"word\"][-3:]}',\n",
        "      f'word[-2:]={row[\"word\"][-2:]}',\n",
        "      f'word.isupper={row[\"word\"].isupper()}',\n",
        "      f'word.isdigit={row[\"word\"].isdigit()}',\n",
        "      f'word.istitle={row[\"word\"].istitle()}',\n",
        "      f'word.isalpha={row[\"word\"].isalpha()}',\n",
        "      f'word.dep={row[\"dep\"]}',\n",
        "      f'word.is_stop={row[\"is_stop\"]}',\n",
        "      f'word.tag:{row[\"tag\"]}',\n",
        "      f'word.lemma:{row[\"lemma\"]}',\n",
        "      f'word.pos:{row[\"pos\"]}',\n",
        "      f'isnounpropn={row[\"pos\"] in [\"NOUN\", \"PROPN\"]}'\n",
        "  ]\n",
        "\n",
        "  if (row.position > 0):\n",
        "    prev_word = data.loc[row.name-1]\n",
        "    features.extend([\n",
        "        f'prev_word.lower={prev_word[\"word\"].lower()}',\n",
        "        f'prev_word.isupper={prev_word[\"word\"].isupper()}',\n",
        "        f'prev_word.isdigit={prev_word[\"word\"].isdigit()}',\n",
        "        f'prev_word.dep={prev_word[\"dep\"]}',\n",
        "        f'prev_word.is_stop={prev_word[\"is_stop\"]}',\n",
        "        f'prev_word.tag:{prev_word[\"tag\"]}',\n",
        "        f'prev_word.lemma:{prev_word[\"lemma\"]}',\n",
        "        f'prevpos:{prev_word[\"pos\"]}'\n",
        "    ])\n",
        "    if(row.position < max(data.position)):\n",
        "      next_word = data.loc[row.name+1]\n",
        "      features.extend([\n",
        "          f'next_word.lower={next_word[\"word\"].lower()}',\n",
        "          f'next_word.isupper={next_word[\"word\"].isupper()}',\n",
        "          f'next_word.isdigit={next_word[\"word\"].isdigit()}',\n",
        "          f'next_word.dep={next_word[\"dep\"]}',\n",
        "          f'next_word.is_stop={next_word[\"is_stop\"]}',\n",
        "          f'next_word.tag:{next_word[\"tag\"]}',\n",
        "          f'next_word.lemma:{next_word[\"lemma\"]}',\n",
        "          f'nextpos:{next_word[\"pos\"]}'\n",
        "      ])\n",
        "  else:\n",
        "    features.append('BEG') # feature to track begin of sentence \n",
        "\n",
        "  if(row.position == len(row.sent)-1):\n",
        "    features.append('END') # feature to track end of sentence\n",
        "\n",
        "  return features"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "haBEOsZGKhb9"
      },
      "source": [
        "## Getting the features"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "Fwb8YUdlNcG1"
      },
      "source": [
        "### Write a code/function to get the features and labels for a sentence"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "emBIm5uK7vtK",
        "notebookRunGroups": {
          "groupValue": "2"
        }
      },
      "outputs": [],
      "source": [
        "# function to take a dataframe and return a list of features and labels\n",
        "def getFeaturesAndLabels(data):\n",
        "  features, labels = [], []\n",
        "  for i in data.sent.unique():\n",
        "    temp_data = data[data.sent == i]\n",
        "    features_temp, labels_temp = [], []\n",
        "    for index, row in temp_data.iterrows():\n",
        "      features_temp.append(getFeaturesForOneWord(row,temp_data))\n",
        "      labels_temp.append(row.label)\n",
        "    features.append(features_temp)\n",
        "    labels.append(labels_temp)\n",
        "  return features, labels"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QG6hYY6rNjS8"
      },
      "source": [
        "## Define input and target variables\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DZmjKbXgN44o"
      },
      "source": [
        "Correctly computing X and Y sequence matrices for training and test data.\n",
        "Check that both sentences and labels are processed"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "U1C4djFXNpqN"
      },
      "source": [
        "### Define the features' values for each sentence as input variable  for CRF model in test and the train dataset "
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "q88-s6BQNr0M"
      },
      "source": [
        "### Define the labels as the target variable for test and the train dataset"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "id": "8Nuz-MjW8LWm",
        "notebookRunGroups": {
          "groupValue": "2"
        }
      },
      "outputs": [],
      "source": [
        "# Get features and labels for train and test data\n",
        "X_train, y_train = getFeaturesAndLabels(train_data)\n",
        "X_test, y_test = getFeaturesAndLabels(test_data)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "notebookRunGroups": {
          "groupValue": "2"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "[['word:All', 'wordword.lower=all', 'word[-3:]=All', 'word[-2:]=ll', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=True', 'word.isalpha=True', 'word.dep=det', 'word.is_stop=True', 'word.tag:DT', 'word.lemma:all', 'word.pos:DET', 'isnounpropn=False', 'BEG'], ['word:live', 'wordword.lower=live', 'word[-3:]=ive', 'word[-2:]=ve', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=amod', 'word.is_stop=False', 'word.tag:JJ', 'word.lemma:live', 'word.pos:ADJ', 'isnounpropn=False', 'prev_word.lower=all', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=det', 'prev_word.is_stop=True', 'prev_word.tag:DT', 'prev_word.lemma:all', 'prevpos:DET', 'next_word.lower=births', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=nsubjpass', 'next_word.is_stop=False', 'next_word.tag:NNS', 'next_word.lemma:birth', 'nextpos:NOUN'], ['word:births', 'wordword.lower=births', 'word[-3:]=ths', 'word[-2:]=hs', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=nsubjpass', 'word.is_stop=False', 'word.tag:NNS', 'word.lemma:birth', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=live', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=amod', 'prev_word.is_stop=False', 'prev_word.tag:JJ', 'prev_word.lemma:live', 'prevpos:ADJ', 'next_word.lower=or', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=cc', 'next_word.is_stop=True', 'next_word.tag:CC', 'next_word.lemma:or', 'nextpos:CCONJ'], ['word:or', 'wordword.lower=or', 'word[-3:]=or', 'word[-2:]=or', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=cc', 'word.is_stop=True', 'word.tag:CC', 'word.lemma:or', 'word.pos:CCONJ', 'isnounpropn=False', 'prev_word.lower=births', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=nsubjpass', 'prev_word.is_stop=False', 'prev_word.tag:NNS', 'prev_word.lemma:birth', 'prevpos:NOUN', 'next_word.lower=23', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=nummod', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:23', 'nextpos:NUM'], ['word:23', 'wordword.lower=23', 'word[-3:]=23', 'word[-2:]=23', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=nummod', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:23', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=or', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=cc', 'prev_word.is_stop=True', 'prev_word.tag:CC', 'prev_word.lemma:or', 'prevpos:CCONJ', 'next_word.lower=weeks', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=conj', 'next_word.is_stop=False', 'next_word.tag:NNS', 'next_word.lemma:week', 'nextpos:NOUN'], ['word:weeks', 'wordword.lower=weeks', 'word[-3:]=eks', 'word[-2:]=ks', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=conj', 'word.is_stop=False', 'word.tag:NNS', 'word.lemma:week', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=23', 'prev_word.isupper=False', 'prev_word.isdigit=True', 'prev_word.dep=nummod', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:23', 'prevpos:NUM', 'next_word.lower=at', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:at', 'nextpos:ADP'], ['word:at', 'wordword.lower=at', 'word[-3:]=at', 'word[-2:]=at', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:at', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=weeks', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=conj', 'prev_word.is_stop=False', 'prev_word.tag:NNS', 'prev_word.lemma:week', 'prevpos:NOUN', 'next_word.lower=the', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=det', 'next_word.is_stop=True', 'next_word.tag:DT', 'next_word.lemma:the', 'nextpos:DET'], ['word:the', 'wordword.lower=the', 'word[-3:]=the', 'word[-2:]=he', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=det', 'word.is_stop=True', 'word.tag:DT', 'word.lemma:the', 'word.pos:DET', 'isnounpropn=False', 'prev_word.lower=at', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:at', 'prevpos:ADP', 'next_word.lower=university', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:NNP', 'next_word.lemma:University', 'nextpos:PROPN'], ['word:University', 'wordword.lower=university', 'word[-3:]=ity', 'word[-2:]=ty', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=True', 'word.isalpha=True', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:NNP', 'word.lemma:University', 'word.pos:PROPN', 'isnounpropn=True', 'prev_word.lower=the', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=det', 'prev_word.is_stop=True', 'prev_word.tag:DT', 'prev_word.lemma:the', 'prevpos:DET', 'next_word.lower=of', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:of', 'nextpos:ADP'], ['word:of', 'wordword.lower=of', 'word[-3:]=of', 'word[-2:]=of', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:of', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=university', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:NNP', 'prev_word.lemma:University', 'prevpos:PROPN', 'next_word.lower=vermont', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:NNP', 'next_word.lemma:Vermont', 'nextpos:PROPN'], ['word:Vermont', 'wordword.lower=vermont', 'word[-3:]=ont', 'word[-2:]=nt', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=True', 'word.isalpha=True', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:NNP', 'word.lemma:Vermont', 'word.pos:PROPN', 'isnounpropn=True', 'prev_word.lower=of', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:of', 'prevpos:ADP', 'next_word.lower=in', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:in', 'nextpos:ADP'], ['word:in', 'wordword.lower=in', 'word[-3:]=in', 'word[-2:]=in', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:in', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=vermont', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:NNP', 'prev_word.lemma:Vermont', 'prevpos:PROPN', 'next_word.lower=1995', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=nummod', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:1995', 'nextpos:NUM'], ['word:1995', 'wordword.lower=1995', 'word[-3:]=995', 'word[-2:]=95', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=nummod', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:1995', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=in', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:in', 'prevpos:ADP', 'next_word.lower=n', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=nummod', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:n', 'nextpos:NUM'], ['word:n', 'wordword.lower=n', 'word[-3:]=n', 'word[-2:]=n', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=nummod', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:n', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=1995', 'prev_word.isupper=False', 'prev_word.isdigit=True', 'prev_word.dep=nummod', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:1995', 'prevpos:NUM', 'next_word.lower=2395', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:2395', 'nextpos:NUM'], ['word:2395', 'wordword.lower=2395', 'word[-3:]=395', 'word[-2:]=95', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:2395', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=n', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=nummod', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:n', 'prevpos:NUM', 'next_word.lower=were', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=auxpass', 'next_word.is_stop=True', 'next_word.tag:VBD', 'next_word.lemma:be', 'nextpos:AUX'], ['word:were', 'wordword.lower=were', 'word[-3:]=ere', 'word[-2:]=re', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=auxpass', 'word.is_stop=True', 'word.tag:VBD', 'word.lemma:be', 'word.pos:AUX', 'isnounpropn=False', 'prev_word.lower=2395', 'prev_word.isupper=False', 'prev_word.isdigit=True', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:2395', 'prevpos:NUM', 'next_word.lower=retrospectively', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=advmod', 'next_word.is_stop=False', 'next_word.tag:RB', 'next_word.lemma:retrospectively', 'nextpos:ADV'], ['word:retrospectively', 'wordword.lower=retrospectively', 'word[-3:]=ely', 'word[-2:]=ly', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=advmod', 'word.is_stop=False', 'word.tag:RB', 'word.lemma:retrospectively', 'word.pos:ADV', 'isnounpropn=False', 'prev_word.lower=were', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=auxpass', 'prev_word.is_stop=True', 'prev_word.tag:VBD', 'prev_word.lemma:be', 'prevpos:AUX', 'next_word.lower=analyzed', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=ROOT', 'next_word.is_stop=False', 'next_word.tag:VBN', 'next_word.lemma:analyze', 'nextpos:VERB'], ['word:analyzed', 'wordword.lower=analyzed', 'word[-3:]=zed', 'word[-2:]=ed', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=ROOT', 'word.is_stop=False', 'word.tag:VBN', 'word.lemma:analyze', 'word.pos:VERB', 'isnounpropn=False', 'prev_word.lower=retrospectively', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=advmod', 'prev_word.is_stop=False', 'prev_word.tag:RB', 'prev_word.lemma:retrospectively', 'prevpos:ADV', 'next_word.lower=for', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:for', 'nextpos:ADP'], ['word:for', 'wordword.lower=for', 'word[-3:]=for', 'word[-2:]=or', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:for', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=analyzed', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=ROOT', 'prev_word.is_stop=False', 'prev_word.tag:VBN', 'prev_word.lemma:analyze', 'prevpos:VERB', 'next_word.lower=delivery', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=compound', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:delivery', 'nextpos:NOUN'], ['word:delivery', 'wordword.lower=delivery', 'word[-3:]=ery', 'word[-2:]=ry', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=compound', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:delivery', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=for', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:for', 'prevpos:ADP', 'next_word.lower=route', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=compound', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:route', 'nextpos:NOUN'], ['word:route', 'wordword.lower=route', 'word[-3:]=ute', 'word[-2:]=te', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=compound', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:route', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=delivery', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=compound', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:delivery', 'prevpos:NOUN', 'next_word.lower=indication', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:indication', 'nextpos:NOUN'], ['word:indication', 'wordword.lower=indication', 'word[-3:]=ion', 'word[-2:]=on', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:indication', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=route', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=compound', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:route', 'prevpos:NOUN', 'next_word.lower=for', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:for', 'nextpos:ADP'], ['word:for', 'wordword.lower=for', 'word[-3:]=for', 'word[-2:]=or', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:for', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=indication', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:indication', 'prevpos:NOUN', 'next_word.lower=cesarean', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=amod', 'next_word.is_stop=False', 'next_word.tag:JJ', 'next_word.lemma:cesarean', 'nextpos:ADJ'], ['word:cesarean', 'wordword.lower=cesarean', 'word[-3:]=ean', 'word[-2:]=an', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=amod', 'word.is_stop=False', 'word.tag:JJ', 'word.lemma:cesarean', 'word.pos:ADJ', 'isnounpropn=False', 'prev_word.lower=for', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:for', 'prevpos:ADP', 'next_word.lower=gestational', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=amod', 'next_word.is_stop=False', 'next_word.tag:JJ', 'next_word.lemma:gestational', 'nextpos:ADJ'], ['word:gestational', 'wordword.lower=gestational', 'word[-3:]=nal', 'word[-2:]=al', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=amod', 'word.is_stop=False', 'word.tag:JJ', 'word.lemma:gestational', 'word.pos:ADJ', 'isnounpropn=False', 'prev_word.lower=cesarean', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=amod', 'prev_word.is_stop=False', 'prev_word.tag:JJ', 'prev_word.lemma:cesarean', 'prevpos:ADJ', 'next_word.lower=age', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=compound', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:age', 'nextpos:NOUN'], ['word:age', 'wordword.lower=age', 'word[-3:]=age', 'word[-2:]=ge', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=compound', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:age', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=gestational', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=amod', 'prev_word.is_stop=False', 'prev_word.tag:JJ', 'prev_word.lemma:gestational', 'prevpos:ADJ', 'next_word.lower=parity', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:parity', 'nextpos:NOUN'], ['word:parity', 'wordword.lower=parity', 'word[-3:]=ity', 'word[-2:]=ty', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:parity', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=age', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=compound', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:age', 'prevpos:NOUN', 'next_word.lower=and', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=cc', 'next_word.is_stop=True', 'next_word.tag:CC', 'next_word.lemma:and', 'nextpos:CCONJ'], ['word:and', 'wordword.lower=and', 'word[-3:]=and', 'word[-2:]=nd', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=cc', 'word.is_stop=True', 'word.tag:CC', 'word.lemma:and', 'word.pos:CCONJ', 'isnounpropn=False', 'prev_word.lower=parity', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:parity', 'prevpos:NOUN', 'next_word.lower=practice', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=compound', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:practice', 'nextpos:NOUN'], ['word:practice', 'wordword.lower=practice', 'word[-3:]=ice', 'word[-2:]=ce', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=compound', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:practice', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=and', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=cc', 'prev_word.is_stop=True', 'prev_word.tag:CC', 'prev_word.lemma:and', 'prevpos:CCONJ', 'next_word.lower=group', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=conj', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:group', 'nextpos:NOUN'], ['word:group', 'wordword.lower=group', 'word[-3:]=oup', 'word[-2:]=up', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=conj', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:group', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=practice', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=compound', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:practice', 'prevpos:NOUN', 'next_word.lower=to', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=aux', 'next_word.is_stop=True', 'next_word.tag:TO', 'next_word.lemma:to', 'nextpos:PART'], ['word:to', 'wordword.lower=to', 'word[-3:]=to', 'word[-2:]=to', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=aux', 'word.is_stop=True', 'word.tag:TO', 'word.lemma:to', 'word.pos:PART', 'isnounpropn=False', 'prev_word.lower=group', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=conj', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:group', 'prevpos:NOUN', 'next_word.lower=reflect', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=advcl', 'next_word.is_stop=False', 'next_word.tag:VB', 'next_word.lemma:reflect', 'nextpos:VERB'], ['word:reflect', 'wordword.lower=reflect', 'word[-3:]=ect', 'word[-2:]=ct', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=advcl', 'word.is_stop=False', 'word.tag:VB', 'word.lemma:reflect', 'word.pos:VERB', 'isnounpropn=False', 'prev_word.lower=to', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=aux', 'prev_word.is_stop=True', 'prev_word.tag:TO', 'prev_word.lemma:to', 'prevpos:PART', 'next_word.lower=risk', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=compound', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:risk', 'nextpos:NOUN'], ['word:risk', 'wordword.lower=risk', 'word[-3:]=isk', 'word[-2:]=sk', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=compound', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:risk', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=reflect', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=advcl', 'prev_word.is_stop=False', 'prev_word.tag:VB', 'prev_word.lemma:reflect', 'prevpos:VERB', 'next_word.lower=status', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=dobj', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:status', 'nextpos:NOUN'], ['word:status', 'wordword.lower=status', 'word[-3:]=tus', 'word[-2:]=us', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=dobj', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:status', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=risk', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=compound', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:risk', 'prevpos:NOUN']] ['O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O']\n",
            "[['word:Furthermore', 'wordword.lower=furthermore', 'word[-3:]=ore', 'word[-2:]=re', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=True', 'word.isalpha=True', 'word.dep=advmod', 'word.is_stop=False', 'word.tag:RB', 'word.lemma:furthermore', 'word.pos:ADV', 'isnounpropn=False', 'BEG'], ['word:when', 'wordword.lower=when', 'word[-3:]=hen', 'word[-2:]=en', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=advmod', 'word.is_stop=True', 'word.tag:WRB', 'word.lemma:when', 'word.pos:SCONJ', 'isnounpropn=False', 'prev_word.lower=furthermore', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=advmod', 'prev_word.is_stop=False', 'prev_word.tag:RB', 'prev_word.lemma:furthermore', 'prevpos:ADV', 'next_word.lower=all', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=det', 'next_word.is_stop=True', 'next_word.tag:DT', 'next_word.lemma:all', 'nextpos:DET'], ['word:all', 'wordword.lower=all', 'word[-3:]=all', 'word[-2:]=ll', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=det', 'word.is_stop=True', 'word.tag:DT', 'word.lemma:all', 'word.pos:DET', 'isnounpropn=False', 'prev_word.lower=when', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=advmod', 'prev_word.is_stop=True', 'prev_word.tag:WRB', 'prev_word.lemma:when', 'prevpos:SCONJ', 'next_word.lower=deliveries', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=nsubjpass', 'next_word.is_stop=False', 'next_word.tag:NNS', 'next_word.lemma:delivery', 'nextpos:NOUN'], ['word:deliveries', 'wordword.lower=deliveries', 'word[-3:]=ies', 'word[-2:]=es', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=nsubjpass', 'word.is_stop=False', 'word.tag:NNS', 'word.lemma:delivery', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=all', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=det', 'prev_word.is_stop=True', 'prev_word.tag:DT', 'prev_word.lemma:all', 'prevpos:DET', 'next_word.lower=were', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=auxpass', 'next_word.is_stop=True', 'next_word.tag:VBD', 'next_word.lemma:be', 'nextpos:AUX'], ['word:were', 'wordword.lower=were', 'word[-3:]=ere', 'word[-2:]=re', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=auxpass', 'word.is_stop=True', 'word.tag:VBD', 'word.lemma:be', 'word.pos:AUX', 'isnounpropn=False', 'prev_word.lower=deliveries', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=nsubjpass', 'prev_word.is_stop=False', 'prev_word.tag:NNS', 'prev_word.lemma:delivery', 'prevpos:NOUN', 'next_word.lower=analyzed', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=advcl', 'next_word.is_stop=False', 'next_word.tag:VBN', 'next_word.lemma:analyze', 'nextpos:VERB'], ['word:analyzed', 'wordword.lower=analyzed', 'word[-3:]=zed', 'word[-2:]=ed', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=advcl', 'word.is_stop=False', 'word.tag:VBN', 'word.lemma:analyze', 'word.pos:VERB', 'isnounpropn=False', 'prev_word.lower=were', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=auxpass', 'prev_word.is_stop=True', 'prev_word.tag:VBD', 'prev_word.lemma:be', 'prevpos:AUX', 'next_word.lower=regardless', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=advmod', 'next_word.is_stop=False', 'next_word.tag:RB', 'next_word.lemma:regardless', 'nextpos:ADV'], ['word:regardless', 'wordword.lower=regardless', 'word[-3:]=ess', 'word[-2:]=ss', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=advmod', 'word.is_stop=False', 'word.tag:RB', 'word.lemma:regardless', 'word.pos:ADV', 'isnounpropn=False', 'prev_word.lower=analyzed', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=advcl', 'prev_word.is_stop=False', 'prev_word.tag:VBN', 'prev_word.lemma:analyze', 'prevpos:VERB', 'next_word.lower=of', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:of', 'nextpos:ADP'], ['word:of', 'wordword.lower=of', 'word[-3:]=of', 'word[-2:]=of', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:of', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=regardless', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=advmod', 'prev_word.is_stop=False', 'prev_word.tag:RB', 'prev_word.lemma:regardless', 'prevpos:ADV', 'next_word.lower=risk', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=compound', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:risk', 'nextpos:NOUN'], ['word:risk', 'wordword.lower=risk', 'word[-3:]=isk', 'word[-2:]=sk', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=compound', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:risk', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=of', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:of', 'prevpos:ADP', 'next_word.lower=status', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:status', 'nextpos:NOUN'], ['word:status', 'wordword.lower=status', 'word[-3:]=tus', 'word[-2:]=us', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:status', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=risk', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=compound', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:risk', 'prevpos:NOUN', 'next_word.lower=but', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=cc', 'next_word.is_stop=True', 'next_word.tag:CC', 'next_word.lemma:but', 'nextpos:CCONJ'], ['word:but', 'wordword.lower=but', 'word[-3:]=but', 'word[-2:]=ut', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=cc', 'word.is_stop=True', 'word.tag:CC', 'word.lemma:but', 'word.pos:CCONJ', 'isnounpropn=False', 'prev_word.lower=status', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:status', 'prevpos:NOUN', 'next_word.lower=limited', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=conj', 'next_word.is_stop=False', 'next_word.tag:VBN', 'next_word.lemma:limit', 'nextpos:VERB'], ['word:limited', 'wordword.lower=limited', 'word[-3:]=ted', 'word[-2:]=ed', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=conj', 'word.is_stop=False', 'word.tag:VBN', 'word.lemma:limit', 'word.pos:VERB', 'isnounpropn=False', 'prev_word.lower=but', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=cc', 'prev_word.is_stop=True', 'prev_word.tag:CC', 'prev_word.lemma:but', 'prevpos:CCONJ', 'next_word.lower=to', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:to', 'nextpos:ADP'], ['word:to', 'wordword.lower=to', 'word[-3:]=to', 'word[-2:]=to', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:to', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=limited', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=conj', 'prev_word.is_stop=False', 'prev_word.tag:VBN', 'prev_word.lemma:limit', 'prevpos:VERB', 'next_word.lower=gestational', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=amod', 'next_word.is_stop=False', 'next_word.tag:JJ', 'next_word.lemma:gestational', 'nextpos:ADJ'], ['word:gestational', 'wordword.lower=gestational', 'word[-3:]=nal', 'word[-2:]=al', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=amod', 'word.is_stop=False', 'word.tag:JJ', 'word.lemma:gestational', 'word.pos:ADJ', 'isnounpropn=False', 'prev_word.lower=to', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:to', 'prevpos:ADP', 'next_word.lower=age', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:NN', 'next_word.lemma:age', 'nextpos:NOUN'], ['word:age', 'wordword.lower=age', 'word[-3:]=age', 'word[-2:]=ge', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:NN', 'word.lemma:age', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=gestational', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=amod', 'prev_word.is_stop=False', 'prev_word.tag:JJ', 'prev_word.lemma:gestational', 'prevpos:ADJ', 'next_word.lower=or', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=cc', 'next_word.is_stop=True', 'next_word.tag:CC', 'next_word.lemma:or', 'nextpos:CCONJ'], ['word:or', 'wordword.lower=or', 'word[-3:]=or', 'word[-2:]=or', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=cc', 'word.is_stop=True', 'word.tag:CC', 'word.lemma:or', 'word.pos:CCONJ', 'isnounpropn=False', 'prev_word.lower=age', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:NN', 'prev_word.lemma:age', 'prevpos:NOUN', 'next_word.lower=36', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=nummod', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:36', 'nextpos:NUM'], ['word:36', 'wordword.lower=36', 'word[-3:]=36', 'word[-2:]=36', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=nummod', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:36', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=or', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=cc', 'prev_word.is_stop=True', 'prev_word.tag:CC', 'prev_word.lemma:or', 'prevpos:CCONJ', 'next_word.lower=weeks', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=conj', 'next_word.is_stop=False', 'next_word.tag:NNS', 'next_word.lemma:week', 'nextpos:NOUN'], ['word:weeks', 'wordword.lower=weeks', 'word[-3:]=eks', 'word[-2:]=ks', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=conj', 'word.is_stop=False', 'word.tag:NNS', 'word.lemma:week', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=36', 'prev_word.isupper=False', 'prev_word.isdigit=True', 'prev_word.dep=nummod', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:36', 'prevpos:NUM', 'next_word.lower=the', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=det', 'next_word.is_stop=True', 'next_word.tag:DT', 'next_word.lemma:the', 'nextpos:DET'], ['word:the', 'wordword.lower=the', 'word[-3:]=the', 'word[-2:]=he', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=det', 'word.is_stop=True', 'word.tag:DT', 'word.lemma:the', 'word.pos:DET', 'isnounpropn=False', 'prev_word.lower=weeks', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=conj', 'prev_word.is_stop=False', 'prev_word.tag:NNS', 'prev_word.lemma:week', 'prevpos:NOUN', 'next_word.lower=rates', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=nsubj', 'next_word.is_stop=False', 'next_word.tag:NNS', 'next_word.lemma:rate', 'nextpos:NOUN'], ['word:rates', 'wordword.lower=rates', 'word[-3:]=tes', 'word[-2:]=es', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=nsubj', 'word.is_stop=False', 'word.tag:NNS', 'word.lemma:rate', 'word.pos:NOUN', 'isnounpropn=True', 'prev_word.lower=the', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=det', 'prev_word.is_stop=True', 'prev_word.tag:DT', 'prev_word.lemma:the', 'prevpos:DET', 'next_word.lower=did', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=aux', 'next_word.is_stop=True', 'next_word.tag:VBD', 'next_word.lemma:do', 'nextpos:AUX'], ['word:did', 'wordword.lower=did', 'word[-3:]=did', 'word[-2:]=id', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=aux', 'word.is_stop=True', 'word.tag:VBD', 'word.lemma:do', 'word.pos:AUX', 'isnounpropn=False', 'prev_word.lower=rates', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=nsubj', 'prev_word.is_stop=False', 'prev_word.tag:NNS', 'prev_word.lemma:rate', 'prevpos:NOUN', 'next_word.lower=not', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=neg', 'next_word.is_stop=True', 'next_word.tag:RB', 'next_word.lemma:not', 'nextpos:PART'], ['word:not', 'wordword.lower=not', 'word[-3:]=not', 'word[-2:]=ot', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=neg', 'word.is_stop=True', 'word.tag:RB', 'word.lemma:not', 'word.pos:PART', 'isnounpropn=False', 'prev_word.lower=did', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=aux', 'prev_word.is_stop=True', 'prev_word.tag:VBD', 'prev_word.lemma:do', 'prevpos:AUX', 'next_word.lower=change', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=ROOT', 'next_word.is_stop=False', 'next_word.tag:VB', 'next_word.lemma:change', 'nextpos:VERB'], ['word:change', 'wordword.lower=change', 'word[-3:]=nge', 'word[-2:]=ge', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=ROOT', 'word.is_stop=False', 'word.tag:VB', 'word.lemma:change', 'word.pos:VERB', 'isnounpropn=False', 'prev_word.lower=not', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=neg', 'prev_word.is_stop=True', 'prev_word.tag:RB', 'prev_word.lemma:not', 'prevpos:PART', 'next_word.lower=12.6', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=nummod', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:12.6', 'nextpos:NUM'], ['word:12.6', 'wordword.lower=12.6', 'word[-3:]=2.6', 'word[-2:]=.6', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=False', 'word.dep=nummod', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:12.6', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=change', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=ROOT', 'prev_word.is_stop=False', 'prev_word.tag:VB', 'prev_word.lemma:change', 'prevpos:VERB', 'next_word.lower=280', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=dobj', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:280', 'nextpos:NUM'], ['word:280', 'wordword.lower=280', 'word[-3:]=280', 'word[-2:]=80', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=dobj', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:280', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=12.6', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=nummod', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:12.6', 'prevpos:NUM', 'next_word.lower=of', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:of', 'nextpos:ADP'], ['word:of', 'wordword.lower=of', 'word[-3:]=of', 'word[-2:]=of', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:of', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=280', 'prev_word.isupper=False', 'prev_word.isdigit=True', 'prev_word.dep=dobj', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:280', 'prevpos:NUM', 'next_word.lower=2214', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:2214', 'nextpos:NUM'], ['word:2214', 'wordword.lower=2214', 'word[-3:]=214', 'word[-2:]=14', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:2214', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=of', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:of', 'prevpos:ADP', 'next_word.lower=primary', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=amod', 'next_word.is_stop=False', 'next_word.tag:JJ', 'next_word.lemma:primary', 'nextpos:ADJ'], ['word:primary', 'wordword.lower=primary', 'word[-3:]=ary', 'word[-2:]=ry', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=amod', 'word.is_stop=False', 'word.tag:JJ', 'word.lemma:primary', 'word.pos:ADJ', 'isnounpropn=False', 'prev_word.lower=2214', 'prev_word.isupper=False', 'prev_word.isdigit=True', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:2214', 'prevpos:NUM', 'next_word.lower=9.2', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=compound', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:9.2', 'nextpos:NUM'], ['word:9.2', 'wordword.lower=9.2', 'word[-3:]=9.2', 'word[-2:]=.2', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=False', 'word.dep=compound', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:9.2', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=primary', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=amod', 'prev_word.is_stop=False', 'prev_word.tag:JJ', 'prev_word.lemma:primary', 'prevpos:ADJ', 'next_word.lower=183', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:183', 'nextpos:NUM'], ['word:183', 'wordword.lower=183', 'word[-3:]=183', 'word[-2:]=83', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:183', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=9.2', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=compound', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:9.2', 'prevpos:NUM', 'next_word.lower=of', 'next_word.isupper=False', 'next_word.isdigit=False', 'next_word.dep=prep', 'next_word.is_stop=True', 'next_word.tag:IN', 'next_word.lemma:of', 'nextpos:ADP'], ['word:of', 'wordword.lower=of', 'word[-3:]=of', 'word[-2:]=of', 'word.isupper=False', 'word.isdigit=False', 'word.istitle=False', 'word.isalpha=True', 'word.dep=prep', 'word.is_stop=True', 'word.tag:IN', 'word.lemma:of', 'word.pos:ADP', 'isnounpropn=False', 'prev_word.lower=183', 'prev_word.isupper=False', 'prev_word.isdigit=True', 'prev_word.dep=pobj', 'prev_word.is_stop=False', 'prev_word.tag:CD', 'prev_word.lemma:183', 'prevpos:NUM', 'next_word.lower=1994', 'next_word.isupper=False', 'next_word.isdigit=True', 'next_word.dep=pobj', 'next_word.is_stop=False', 'next_word.tag:CD', 'next_word.lemma:1994', 'nextpos:NUM'], ['word:1994', 'wordword.lower=1994', 'word[-3:]=994', 'word[-2:]=94', 'word.isupper=False', 'word.isdigit=True', 'word.istitle=False', 'word.isalpha=False', 'word.dep=pobj', 'word.is_stop=False', 'word.tag:CD', 'word.lemma:1994', 'word.pos:NUM', 'isnounpropn=False', 'prev_word.lower=of', 'prev_word.isupper=False', 'prev_word.isdigit=False', 'prev_word.dep=prep', 'prev_word.is_stop=True', 'prev_word.tag:IN', 'prev_word.lemma:of', 'prevpos:ADP']] ['O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O', 'O']\n"
          ]
        }
      ],
      "source": [
        "# printing the features and labels for example train and test data\n",
        "print(X_train[0],y_train[0])\n",
        "print(X_test[0],y_test[0])"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "YrIAz8gqN_Js"
      },
      "source": [
        "## Build the CRF Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 19,
      "metadata": {
        "id": "k2oGSiqQ8Otd",
        "notebookRunGroups": {
          "groupValue": "2"
        }
      },
      "outputs": [],
      "source": [
        "# Build the CRF model.\n",
        "crf=sklearn_crfsuite.CRF(max_iterations=200,all_possible_transitions=True)\n",
        "try:\n",
        "  crf.fit(X_train, y_train)\n",
        "except:\n",
        "  print('Error in training')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5VPFrPZJOJMq"
      },
      "source": [
        "## Evaluation"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "84gUagMZOP9B"
      },
      "source": [
        "### Predict the labels of each of the tokens in each sentence of the test dataset that has been pre processed earlier."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 20,
      "metadata": {
        "id": "D_sDJs1n8-sI",
        "notebookRunGroups": {
          "groupValue": "2"
        }
      },
      "outputs": [],
      "source": [
        "Y_pred = crf.predict(X_test)\n",
        "temp_df = test_data[['sent','word','label']]\n",
        "temp_df['pred'] = [item for sublist in Y_pred for item in sublist]"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "qIneYrqZOUiJ"
      },
      "source": [
        "### Calculate the f1 score using the actual labels and the predicted labels of the test dataset."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 21,
      "metadata": {
        "id": "Lr_JNIgrOTcr",
        "notebookRunGroups": {
          "groupValue": "2"
        }
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "0.8868022122690995"
            ]
          },
          "execution_count": 21,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "metrics.flat_f1_score(y_test, Y_pred, average='weighted')"
      ]
    },
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "id": "ORDyDw6sYHf5"
      },
      "source": [
        "## Identifying Diseases and Treatments using Custom NER\n",
        "\n",
        "We now use the CRF model's prediction to prepare a record of diseases identified in the corpus and treatments used for the diseases.\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "V9aOa7LEOcWu"
      },
      "source": [
        "### Create the logic to get all the predicted treatments (T) labels corresponding to each disease (D) label in the test dataset."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sWxZ2TaOqN-u"
      },
      "source": [
        "![image.png](data:image/png;base64,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)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 22,
      "metadata": {},
      "outputs": [],
      "source": [
        "disease=[]\n",
        "treatment=[]\n",
        "sentence=[]\n",
        "med_dict = {}\n",
        "for i in temp_df.sent.unique():\n",
        "  try:\n",
        "    val = temp_df.loc[temp_df['sent'] == i,'pred'].value_counts()\n",
        "    if True:\n",
        "      disease_val = temp_df[temp_df['pred'] == 'D'].loc[temp_df['sent'] == i,'word']\n",
        "      treatment_val = temp_df[temp_df['pred'] == 'T'].loc[temp_df['sent'] == i,'word']\n",
        "      disease_single = disease_val if type(disease_val) == str else \" \".join(disease_val)\n",
        "      treatment_single = treatment_val if type(treatment_val) == str else \" \".join(treatment_val)\n",
        "      if disease_single not in disease:\n",
        "          med_dict[disease_single] = treatment_single\n",
        "      else:\n",
        "          print('Entered')\n",
        "          med_dict[disease_single] = med_dict.get(disease_single)+'/'+treatment_single\n",
        "  except AttributeError:\n",
        "    pass"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TvTodSY5OfZ9"
      },
      "source": [
        "### Predict the treatment for the disease name: 'hereditary retinoblastoma'"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "id": "w_TonOcVAG1P"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "{'': 'Intravenous immune globulin', 'gestational diabetes': '', 'hydrops fetalis': '', 'preeclampsia': '', 'asymmetric double hemiplegia': '', 'cancer': 'Matrix metalloproteinase inhibitors', 'breast cancer ovarian cancer': '', 'prostate cancer': 'radical prostatectomy and iodine 125 interstitial radiotherapy', 'multiple sclerosis': 'Intravenous immunoglobulin treatment', 'retinoblastoma': 'radiotherapy', 'epilepsy': 'Methylphenidate', 'coronary -': 'antibiotics', 'early - stage cervical carcinoma': '', 'cerebral palsy': 'Hyperbaric oxygen therapy', 'severe pain': '', 'myofascial trigger point pain': '', 'confirmed infection': '', 'infections': '', 'primary pulmonary hypertension PPH': 'fenfluramines', 'osteoporosis': '', 'Curtis Radiat': '', 'foot infections': '', 'foot infection': '', 'intracranial hemorrhage with the double -': '', 'hypertension': '', 'rheumatoid arthritis': '', 'arthrodesis': 'arthrodesis', 'treatment of early': 'Ropinirole monotherapy', 'sore throat': '', 'Crohn disease': 'steroids', 'treatment of female stress urinary incontinence': '', 'stress urinary incontinence': '', 'corpal gastritis': '', 'preeclampsia proteinuric hypertension': 'intrauterine insemination with donor sperm versus intrauterine insemination with partner sperm', 'mild preeclampsia preeclampsia': '', 'heart failure amyloidosis': '', 'cardiac amyloid heart disease primary AL amyloidosis': '', 'intra -': '', 'major pulmonary embolism': '', 'malignant pleural mesothelioma': '', 'dyskinesias': '', 'hearing loss': '', 'Darier disease': '', 'tumor necrosis factor': 'interferon - gamma', 'hemorrhagic shock TSS': '', 'mediastinal teratomas': 'chemotherapy', 'non -': 'testicular fine needle aspiration TEFNA open biopsy and testicular sperm extraction TESE', 'acute lymphoblastic leukaemias': '', 'hormonal dysfunction': '', 'endometriosis': '', 'diabetes': '', 'hyperlipidemias': '', 'syndrome': 'cholecystectomy in', 'deep vein thrombosis DVT': '', 'The': '', 'Hepatocerebral': '', 'in': '', 'carcinoma challenge and': '', 'haemolytic disease an': '', 'ulcer': 'Measured for', 'schizophrenia': '', 'Polycystic adrenal hyperplasia': '', 'A': '', 'of renal stones': '', 'Diagnosis disease': '', 'Medical disease': '', 'Is': '', 'Aromatization of': '', 'Ovarian carcinoma': '', 'The problems': '', 'in pregnancy': '', 'Fat absorption': '', 'Potassium in': '', 'Clinical physiopathology': '', 'heart congestive failure': '', 'Defects of': '', 'The stimulatory': '', 'increased': '', 'external ophthalmoplegia a case': '', 'leukomalacia in preterm': '', 'laryngeal carcinomas': '', 'A case': '', 'Single -': 'dose pharmacokinetics', 'in the': 'for colorectal', 'Elective surgery': 'for gastrointestinal', 'and attention': 'effective', 'Chlamydia trachomatis': '', 'anemia of end - stage': '', 'malignant lymphomas': '', 'polymyalgia rheumatica and temporal arteritis': '', 'Obstructive sleep apnoea': '', 'coronary heart disease': '', 'zary syndrome': 'Extracorporeal photopheresis', 'Russia': '', 'primary pulmonary hypertension': 'fenfluramine and dexfenfluramine', 'restenosis': 'coronary angioplasty', 'oral Crohn disease refractory': '', 'female androgenetic alopecia': '', 'disconnection syndrome': '', 'Spontaneous splenic rupture': '', 'HIV': '', 'gunshot wounds to the abdomen': '', 'Blastocystis hominis infection epidemiologic study': '', 'primary stomach lymphoma': '', 'systemic lupus erythematosus mite dermatitis': '', 'lung cancer': '', 'Metastatic malignant mesothelioma colonic polyps': '', 'treatment of Parkinson disease postoperative magnetic resonance': 'Microelectrode - guided posteroventral pallidotomy', 'colorectal cancer': 'UFT', 'hepatitis virus strain 3 infection': '', 'tacrolimus therapy in renal transplantation': '', 'acute respiratory distress syndrome': '', 'Homeopathy for postoperative ileus A meta': '', 'chronic hepatitis grading': '', 'Postraumatic Stress Disorder': '', 'Pendred syndrome': '', 'Abstracts': '', 'and hyperlipidemias': '', 'Role': '', 'Golf': '', 'hematomas reduced': '', 'soft tissue sarcomas': '', 'Radiotherapy tissue sarcomas': '', 'Intramedullary metastases of bronchogenic carcinoma': '', 'Nervous': '', 'nerve injuries during': '', 'perception disorders due to bilateral cortical lesions': '', 'Generation': '', 'injuries accompanying': '', 'endothelial dysfunction in': '', 'pregnancy': '', 'intractable seizures': '', 'acute myocardial infarction': '', 'mutation in a': '', 'missense mutation R1623Q': '', 'motherhood can': '', 'plasmacytoid cells arising in': '', 'cardiac surgery': '', 'breast cancer': 'undergone subcutaneous mastectomy', 'malaria': '', 'severe psoriasis': 'systemic treatments', 'leukemia': '', 'Acute obstructive hydrocephalus complicating bacterial meningitis': '', 'non - small cell': 'surgery', 'inoperable advanced malignancies such as colorectal cancer': 'combination with leucovorin or cisplatin', 'advanced nsclc manageable toxicities': '', 'recurrent cervical adenocarcinoma': 'weekly paclitaxel', 'common cold': 'PURPOSE Macrolide antibiotics', 'influenza influenza control': '', 'pain': '', 'pleuritic pain': '', 'fungal infections': '', 'inflammatory skin diseases': 'topical corticosteroids', 'unresectable stage iii nsclc': 'sequential chemotherapy followed by radiation', 'nsclc nsclc sclc limited disease': 'got surgical treatment', 'locally advanced non - small -': '', 'patients with limited stage small cell lung cancer': 'vip combination chemotherapy and early concurrent thoracic irradiation', 'malignant pleural effusions from nsclc': 'systemic chemotherapy', 'small': '', 'intraluminal early - stage': '', 'pathologic stage i non - small': 'uft a 5 - fluorouracil derivative chemotherapeutic agent', 'supraclavicular node metastases in nsclc': '', 'non - small - cell - lung - cancer nsclc': '', 'esophageal and tracheobronchial stenoses esophageal cancer lung cancer': '', 'esophageal and tracheobronchial stenoses': '', 'osteogenic sarcoma': '', 'melanoma renal cell and ovarian carcinoma brain metastasis': '', 'colorectal metastases': '', 'limited - stage small - cell carcinoma of the lung': '', 'nsclc': '', 'metastatic colorectal cancer': 'intravenous oxaliplatin', 'non - hodgkin lymphoma breast cancer mesothelioma and non -': '', 'primary tumor bronchogenic carcinoma': 'resection', 'non - small cell lung cancer': 'paclitaxel and carboplatin', 'primary lung cancer adenocarcinoma ad': 'resection', 'stage iii nsclc': 'chemotherapy surgery', 'small cell lung cancer': '', 'small - cell': 'combination chemotherapy', 'primary cancer': '', 'advanced non small - cell': '', 'untreated small cell lung cancer sclc': 'chemotherapy', 'patients with pulmonary hypertension': '', 'head and neck cancer': 'pretreatment with intravenous amifostine', 'psoriasis': '', 'disseminated malignant melanoma': 'leukocyte A recombinant interferon rIFN -', 'therapy of ventricular tachycardia': '', 'colds': '', 'non - functioning endocrine pancreatic tumor with progressive liver and lymph node metastases': '', 'severe acute hepatitis accompanying cholestasis autoimmune hepatitis': '', 'syringomyelia associated with spinal adhesive arachnoiditis': 'Surgical management', 'biliary colic symptoms biliary dyskinesia': 'cholecystectomy', 'biliary': 'Cholecystectomy', 'paranasal sinuses associated with the common cold': 'pseudoephedrine plus acetaminophen', 'treatment of pulmonary aspergilloma': '', 'viral infections': 'Interferon', 'asthma': '', 'Pneumocystis carinii pneumonia': 'trimethoprim - sulfamethoxazole', 'head lice': '', 'rhinovirus colds rhinovirus': 'clarithromycin', 'chronic bronchitis': '', 'influenza breast cancer': 'vaccination', 'infection': '', 'carcinoma': 'esophagectomy', 'Contemporary asthma persistent asthma': 'corticosteroids', 'hepatitis C viremia': 'combination therapy', 'CBD stones': '', 'bacterial meningitis Haemophilus influenzae meningitis': 'antibiotic - resistant', 'meningitis': 'New vaccines', 'RSV bronchiolitis': '', 'primary sclerosing cholangitis PSC': 'oral budesonide', 'depression': '', 'chronic renal failure': '', 'ischemic heart disease': 'Aortocoronary bypass grafting', 'autologous cancer cancer a reduced metastatic load': '', 'acute occlusion of the middle cerebral artery': '', 'open heart surgery': '', 'benign prostatic hyperplasia BPH': 'conservative pharmacological treatment', 'rheumatic pain': '', 'empyema': '', 'large - bowel cancer': '', 'phaeochromocytoma': 'Adrenalectomy', 'malignant melanoma interferon alfa-2a': '', 'advanced renal': 'various interferon', 'low - grade': 'interferon alpha', 'chronic myelogenous leukemia CML': '', 'esophageal achalasia': '', 'irritable bowel syndrome': '', 'proximal hypospadias': '', 'tumors with autologous tumor - derived heat shock': '', 'sickle cell disease': 'hydroxyurea', 'Haematoma block': '', 'necrosis': '', 'resistant bladder cancer': 'photodynamic therapy in combination with mitomycin C', 'primary uveal melanoma': '', 'advanced rectal cancer': '', 'bronchiectasis': 'Current surgical therapy', 'spontaneous pneumothorax': 'Thoracoscopic surgery', 'noncirrhotic portal hypertension long -': '', 'midline incisional hernias': '', 'acute cerebral ischemia': 'Antiplatelet therapy', 'hyperopia': '', 'Interferon treatment of renal cell carcinoma': '', 'odontogenic infections': '', 'Barrett esophagus': '', 'Frontal dysfunction Alzheimer disease': '', 'sciatica': 'Corticosteroid injections', 'brain hemorrhage atrial fibrillation': '', 'lymphoma': '', 'acute repetitive seizures': '', 'pulmonary hypertension': 'Single or double lung transplantation', 'acoustic neuroma': 'Stereotactic radiosurgery', 'postvitrectomy diabetic vitreous hemorrhage': 'Peripheral retinal cryotherapy', 'acute carbon monoxide poisoning': '', 'pertussis measles rubella': 'vaccines', 'acute colonic pseudo': '', 'severe secondary peritonitis': 'Surgical management', 'hypertension with ascorbic acid': '', 'resection of hepatic metastases from colorectal cancer': 'Hepatic arterial infusion of chemotherapy', 'treatment of severe diverticular hemorrhage': '', 'stable asthma': '', 'arthritis': ''}\n"
          ]
        }
      ],
      "source": [
        "print(med_dict)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "id": "YeNQRLrIYRay"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Treatment not in database\n"
          ]
        }
      ],
      "source": [
        "try:\n",
        "  print(med_dict['hereditary retinoblastoma'])\n",
        "except KeyError:\n",
        "  print('Treatment not in database')"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "name": "Assignment.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "tf",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.0"
    },
    "vscode": {
      "interpreter": {
        "hash": "5b78c356f59ad4f4bb37d320e87edfdd6e4609350ddbfb799f14c15ad9db04c2"
      }
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}