Clustering_Clinical_Trials.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Clustering-Clinical-Trials.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pxHtsvK_kEeo",
        "colab_type": "text"
      },
      "source": [
        "In this notebook, I will be performing some clustering on the COVID-19 trials data available from WHO. Each clinical trial has certain interventions defined as: A process or action that is the focus of a clinical study. Interventions include drugs, medical devices, procedures, vaccines, and other products that are either investigational or already available. Interventions can also include noninvasive approaches, such as education or modifying diet and exercise. \n",
        "\n",
        "Clustering the interventions gives us the ability to get a better view of the nature of these clinical trials which further coupled with 'geolocation data' and 'clinical trials to research mapping' can be extremely insightful"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tpHf0fdcToAK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')\n",
        "#Getting the imports done\n",
        "import numpy as np \n",
        "import pandas as pd \n",
        "import os\n",
        "import re\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf\n",
        "import xml.etree.ElementTree as ET"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QNiGyk8MTq6p",
        "colab_type": "code",
        "outputId": "aef56c92-7c6d-4daa-8190-7871defef4a1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 643
        }
      },
      "source": [
        "#Reading in our file\n",
        "df = pd.read_csv(\"gdrive/My Drive/COVID19.csv\")\n",
        "df.head()"
      ],
      "execution_count": 244,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>TrialID</th>\n",
              "      <th>Last Refreshed on</th>\n",
              "      <th>Public title</th>\n",
              "      <th>Scientific title</th>\n",
              "      <th>Acronym</th>\n",
              "      <th>Primary sponsor</th>\n",
              "      <th>Date registration</th>\n",
              "      <th>Date registration3</th>\n",
              "      <th>Export date</th>\n",
              "      <th>Source Register</th>\n",
              "      <th>web address</th>\n",
              "      <th>Recruitment Status</th>\n",
              "      <th>other records</th>\n",
              "      <th>Inclusion agemin</th>\n",
              "      <th>Inclusion agemax</th>\n",
              "      <th>Inclusion gender</th>\n",
              "      <th>Date enrollement</th>\n",
              "      <th>Target size</th>\n",
              "      <th>Study type</th>\n",
              "      <th>Study design</th>\n",
              "      <th>Phase</th>\n",
              "      <th>Countries</th>\n",
              "      <th>Contact Firstname</th>\n",
              "      <th>Contact Lastname</th>\n",
              "      <th>Contact Address</th>\n",
              "      <th>Contact Email</th>\n",
              "      <th>Contact Tel</th>\n",
              "      <th>Contact Affiliation</th>\n",
              "      <th>Inclusion Criteria</th>\n",
              "      <th>Exclusion Criteria</th>\n",
              "      <th>Condition</th>\n",
              "      <th>Intervention</th>\n",
              "      <th>Primary outcome</th>\n",
              "      <th>results date posted</th>\n",
              "      <th>results date completed</th>\n",
              "      <th>results url link</th>\n",
              "      <th>Retrospective flag</th>\n",
              "      <th>Bridging flag truefalse</th>\n",
              "      <th>Bridged type</th>\n",
              "      <th>results yes no</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>NCT04246242</td>\n",
              "      <td>3 February 2020</td>\n",
              "      <td>A Randomized Multicenter Controlled Clinical T...</td>\n",
              "      <td>A Randomized Multicenter Controlled Clinical T...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Xiangya Hospital of Central South University</td>\n",
              "      <td>27/01/2020</td>\n",
              "      <td>20200127</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ClinicalTrials.gov</td>\n",
              "      <td>https://clinicaltrials.gov/show/NCT04246242</td>\n",
              "      <td>Not recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>All</td>\n",
              "      <td>January 25, 2020</td>\n",
              "      <td>500</td>\n",
              "      <td>Interventional</td>\n",
              "      <td>Allocation: Randomized. Intervention model: Pa...</td>\n",
              "      <td>Phase 4</td>\n",
              "      <td>NaN</td>\n",
              "      <td></td>\n",
              "      <td>Pinhua Pan, Medical PhD</td>\n",
              "      <td>NaN</td>\n",
              "      <td>pinhuapan668@126.com</td>\n",
              "      <td>+86 13574810968</td>\n",
              "      <td>NaN</td>\n",
              "      <td>\\r&lt;br&gt;        Inclusion Criteria:\\r&lt;br&gt;\\r&lt;br&gt; ...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>2019 Novel Coronavirus</td>\n",
              "      <td>Drug: Conventional treatment group;Drug: arbid...</td>\n",
              "      <td>mortality</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>ChiCTR2000029949</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>A Medical Records Based Study for the Effectiv...</td>\n",
              "      <td>A Medical Records Based Study for the Effectiv...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Emergency Department of Zhongnan hospital of W...</td>\n",
              "      <td>2020-02-16</td>\n",
              "      <td>20200216</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ChiCTR</td>\n",
              "      <td>http://www.chictr.org.cn/showproj.aspx?proj=49181</td>\n",
              "      <td>Not Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>18</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Both</td>\n",
              "      <td>2020-02-16</td>\n",
              "      <td>Case series:40;</td>\n",
              "      <td>Observational study</td>\n",
              "      <td>Sequential</td>\n",
              "      <td>Retrospective study</td>\n",
              "      <td>China</td>\n",
              "      <td>Yan, Zhao</td>\n",
              "      <td>NaN</td>\n",
              "      <td>169 Donghu Road, Wuchang District, Wuhan, Hube...</td>\n",
              "      <td>doctoryanzhao@whu.edu.cn</td>\n",
              "      <td>+86 13995577963</td>\n",
              "      <td>Emergency Department of Zhongnan hospital of W...</td>\n",
              "      <td>Inclusion criteria: 1. Aged &gt;=18 years; \\r&lt;br&gt;...</td>\n",
              "      <td>Exclusion criteria: Absolutely forbidden crite...</td>\n",
              "      <td>Novel Coronavirus Pneumonia (COVID-19)</td>\n",
              "      <td>Case series:Extracorporeal Membrane Oxygenation;</td>\n",
              "      <td>inhospital length;inhospital mortality;ECMO tr...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>NCT04245631</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>Development of a Simple, Fast and Portable Rec...</td>\n",
              "      <td>Development of a Simple, Fast and Portable Rec...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Beijing Ditan Hospital</td>\n",
              "      <td>26/01/2020</td>\n",
              "      <td>20200126</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ClinicalTrials.gov</td>\n",
              "      <td>https://clinicaltrials.gov/show/NCT04245631</td>\n",
              "      <td>Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>1 Year</td>\n",
              "      <td>90 Years</td>\n",
              "      <td>All</td>\n",
              "      <td>January 1, 2020</td>\n",
              "      <td>50</td>\n",
              "      <td>Observational</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>China</td>\n",
              "      <td>; ;</td>\n",
              "      <td>Yao Xie, Doctor;Yao Xie, Doctor;Yao Xie, Doctor</td>\n",
              "      <td>NaN</td>\n",
              "      <td>;xieyao00120184@sina.com;xieyao00120184@sina.com</td>\n",
              "      <td>;8610-84322200;8610-84322200</td>\n",
              "      <td>Department of Hepatology, Division 2, Beijing ...</td>\n",
              "      <td>\\r&lt;br&gt;        Inclusion Criteria:\\r&lt;br&gt;\\r&lt;br&gt; ...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>New Coronavirus</td>\n",
              "      <td>Diagnostic Test: Recombinase aided amplificati...</td>\n",
              "      <td>Detection specificity is greater than 95%;Dete...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>NCT04252885</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>The Efficacy of Lopinavir Plus Ritonavir and A...</td>\n",
              "      <td>A Randomized, Open-label, Controlled Study of ...</td>\n",
              "      <td>ELACOI</td>\n",
              "      <td>Guangzhou 8th People's Hospital</td>\n",
              "      <td>30/01/2020</td>\n",
              "      <td>20200130</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ClinicalTrials.gov</td>\n",
              "      <td>https://clinicaltrials.gov/show/NCT04252885</td>\n",
              "      <td>Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>18 Years</td>\n",
              "      <td>80 Years</td>\n",
              "      <td>All</td>\n",
              "      <td>January 28, 2020</td>\n",
              "      <td>125</td>\n",
              "      <td>Interventional</td>\n",
              "      <td>Allocation: Randomized. Intervention model: Pa...</td>\n",
              "      <td>Phase 4</td>\n",
              "      <td>China</td>\n",
              "      <td>;</td>\n",
              "      <td>Yueping Li, MD, Master;Linghua Li, MD, PhD</td>\n",
              "      <td>NaN</td>\n",
              "      <td>lipipi007@163.com;</td>\n",
              "      <td>020- 36052117;</td>\n",
              "      <td>NaN</td>\n",
              "      <td>\\r&lt;br&gt;        Inclusion Criteria:\\r&lt;br&gt;\\r&lt;br&gt; ...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Coronavirus Infections</td>\n",
              "      <td>Drug: Lopinavir and Ritonavir Tablets;Drug: Ar...</td>\n",
              "      <td>The rate of virus inhibition</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>NCT04255940</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>2019-nCoV Outbreak and Cardiovascular Diseases</td>\n",
              "      <td>Impact of a Novel Coronavirus (2019-nCoV) Outb...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Qilu Hospital of Shandong University</td>\n",
              "      <td>03/02/2020</td>\n",
              "      <td>20200203</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ClinicalTrials.gov</td>\n",
              "      <td>https://clinicaltrials.gov/show/NCT04255940</td>\n",
              "      <td>Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>All</td>\n",
              "      <td>January 20, 2020</td>\n",
              "      <td>12000</td>\n",
              "      <td>Observational</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Phase 4</td>\n",
              "      <td>China</td>\n",
              "      <td></td>\n",
              "      <td>Panpan Hao, MD</td>\n",
              "      <td>NaN</td>\n",
              "      <td>panda.how@sdu.edu.cn</td>\n",
              "      <td>86-18560086593</td>\n",
              "      <td>NaN</td>\n",
              "      <td>\\r&lt;br&gt;        Inclusion Criteria:\\r&lt;br&gt;\\r&lt;br&gt; ...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Cardiovascular Death; Major Adverse Cardiovasc...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Cardiovascular Death</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "            TrialID Last Refreshed on  ... Bridged type results yes no\n",
              "0       NCT04246242   3 February 2020  ...                         NaN\n",
              "1  ChiCTR2000029949  17 February 2020  ...                         NaN\n",
              "2       NCT04245631  17 February 2020  ...                         NaN\n",
              "3       NCT04252885  17 February 2020  ...                         NaN\n",
              "4       NCT04255940  17 February 2020  ...                         NaN\n",
              "\n",
              "[5 rows x 40 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 244
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "J5iLX5HNlf6d",
        "colab_type": "text"
      },
      "source": [
        "While cleaning the columns, I observed that since the data source may be scraped, it consists some junk characters and numbers that distort our results. Moreover, the terms \"covid\", \"coronavirus\", \"novel\" and \"ncov\" all are often present and add little value to the intervention type"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Av-6jK2fT1Es",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Cleaning the columns\n",
        "import re\n",
        "def cleaner(data):\n",
        "  data = data.replace('br','')\n",
        "  data = data.replace('covid','')\n",
        "  data = data.replace('coronavirus','')\n",
        "  data = data.replace('ncov','')\n",
        "  data = data.replace('novel','')\n",
        "  data = re.sub(r'([^a-zA-Z ]+?)', ' ', data).lower()\n",
        "  data = re.sub(r\"\\s\\s+\",\" \",data)\n",
        "  return data\n",
        "\n",
        "df = df.dropna(subset=['Intervention'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "idzy_JQOWSRI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#Next step is to prepare a corpus for our TfIdf vectors\n",
        "def build_corpus(data,col):\n",
        "  corpus=[]\n",
        "  for sentence in data[col].iteritems():\n",
        "    corpus.append(sentence[1].rstrip())\n",
        "  return corpus\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6CfdGwb5WUFb",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.feature_extraction.text import TfidfVectorizer\n",
        "tfidf = TfidfVectorizer(stop_words='english')\n",
        "from sklearn.preprocessing import normalize\n",
        "def uniquevals(data):\n",
        "  l = list(set(data.split(\" \")))\n",
        "  return \" \".join(l)\n",
        "#In this short text matching I dont really care about order of words and don't really intend to capture repetitions.\n",
        "#Rather I just want to capture some keywords\n",
        "df['Intervention'] = df['Intervention'].apply(cleaner)\n",
        "df['Intervention'] = df['Intervention'].apply(uniquevals)\n",
        "corpus = build_corpus(df,'Intervention')\n",
        "xvec = tfidf.fit_transform(corpus) \n",
        "tf_idf_norm = normalize(xvec)\n",
        "tf_idf_array = tf_idf_norm.toarray()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0GMEX5RJowjx",
        "colab_type": "text"
      },
      "source": [
        "Our TF-IDF vector will have a high number of dimensions and can suffer from the curse of dimensionality so we reduce it using PCA"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X8-U90vL0hY1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.decomposition import PCA\n",
        "sklearn_pca = PCA(n_components = 2)\n",
        "pca_vec = sklearn_pca.fit_transform(tf_idf_array)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "LwnpOApWo8PV",
        "colab_type": "text"
      },
      "source": [
        "Next, we observe the SSE vs No of clusters graph to observe an elbow as a possible estimate of the number of clusters"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q5vhLgnSnJ9W",
        "colab_type": "code",
        "outputId": "f7be3903-0a98-40d1-dca4-30ddb2c3b945",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 562
        }
      },
      "source": [
        "#K Means Clustering on Interventions\n",
        "#To choose the number of cluster, we use an SSE measure and observe elbows in the graph\n",
        "from sklearn.cluster import KMeans\n",
        "def find_optimal_clusters(data, max_k):\n",
        "    iters = range(2, max_k+1, 2)\n",
        "    sse = []\n",
        "    for k in iters:\n",
        "        sse.append(KMeans(n_clusters=k, random_state=42).fit(data).inertia_)\n",
        "        print('Fit {} clusters'.format(k))\n",
        "        \n",
        "    f, ax = plt.subplots(1, 1)\n",
        "    ax.plot(iters, sse, marker='o')\n",
        "    ax.set_xlabel('Cluster Centers')\n",
        "    ax.set_xticks(iters)\n",
        "    ax.set_xticklabels(iters)\n",
        "    ax.set_ylabel('SSE')\n",
        "    \n",
        "find_optimal_clusters(pca_vec, 30)"
      ],
      "execution_count": 249,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Fit 2 clusters\n",
            "Fit 4 clusters\n",
            "Fit 6 clusters\n",
            "Fit 8 clusters\n",
            "Fit 10 clusters\n",
            "Fit 12 clusters\n",
            "Fit 14 clusters\n",
            "Fit 16 clusters\n",
            "Fit 18 clusters\n",
            "Fit 20 clusters\n",
            "Fit 22 clusters\n",
            "Fit 24 clusters\n",
            "Fit 26 clusters\n",
            "Fit 28 clusters\n",
            "Fit 30 clusters\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-vSOu9WcpKU9",
        "colab_type": "text"
      },
      "source": [
        "At around 6 clusters, we can observe an \"elbow\". We shall use this for our K-Means algorithm. In the next step, I will also create a scatter plot to visualize how our K-Means clustering has worked"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6XmflB9tl9HO",
        "colab_type": "code",
        "outputId": "3fc0b570-22a4-43f0-e049-b98f432e3546",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 265
        }
      },
      "source": [
        "kmeans = KMeans(n_clusters=6)\n",
        "fitted = kmeans.fit(pca_vec)\n",
        "prediction = kmeans.predict(pca_vec)\n",
        "plt.scatter(Y_sklearn[:, 0], Y_sklearn[:, 1],c=prediction, s=50, cmap='viridis')\n",
        "plt.show()"
      ],
      "execution_count": 250,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TsbBMxQK3EN6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 678
        },
        "outputId": "f252150f-10a5-48a0-c576-9e3993e367ff"
      },
      "source": [
        "#Let's add our predicted classes to the dataset and save it\n",
        "df['class_label'] = prediction\n",
        "df.head()"
      ],
      "execution_count": 251,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>TrialID</th>\n",
              "      <th>Last Refreshed on</th>\n",
              "      <th>Public title</th>\n",
              "      <th>Scientific title</th>\n",
              "      <th>Acronym</th>\n",
              "      <th>Primary sponsor</th>\n",
              "      <th>Date registration</th>\n",
              "      <th>Date registration3</th>\n",
              "      <th>Export date</th>\n",
              "      <th>Source Register</th>\n",
              "      <th>web address</th>\n",
              "      <th>Recruitment Status</th>\n",
              "      <th>other records</th>\n",
              "      <th>Inclusion agemin</th>\n",
              "      <th>Inclusion agemax</th>\n",
              "      <th>Inclusion gender</th>\n",
              "      <th>Date enrollement</th>\n",
              "      <th>Target size</th>\n",
              "      <th>Study type</th>\n",
              "      <th>Study design</th>\n",
              "      <th>Phase</th>\n",
              "      <th>Countries</th>\n",
              "      <th>Contact Firstname</th>\n",
              "      <th>Contact Lastname</th>\n",
              "      <th>Contact Address</th>\n",
              "      <th>Contact Email</th>\n",
              "      <th>Contact Tel</th>\n",
              "      <th>Contact Affiliation</th>\n",
              "      <th>Inclusion Criteria</th>\n",
              "      <th>Exclusion Criteria</th>\n",
              "      <th>Condition</th>\n",
              "      <th>Intervention</th>\n",
              "      <th>Primary outcome</th>\n",
              "      <th>results date posted</th>\n",
              "      <th>results date completed</th>\n",
              "      <th>results url link</th>\n",
              "      <th>Retrospective flag</th>\n",
              "      <th>Bridging flag truefalse</th>\n",
              "      <th>Bridged type</th>\n",
              "      <th>results yes no</th>\n",
              "      <th>class_label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>NCT04246242</td>\n",
              "      <td>3 February 2020</td>\n",
              "      <td>A Randomized Multicenter Controlled Clinical T...</td>\n",
              "      <td>A Randomized Multicenter Controlled Clinical T...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Xiangya Hospital of Central South University</td>\n",
              "      <td>27/01/2020</td>\n",
              "      <td>20200127</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ClinicalTrials.gov</td>\n",
              "      <td>https://clinicaltrials.gov/show/NCT04246242</td>\n",
              "      <td>Not recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>All</td>\n",
              "      <td>January 25, 2020</td>\n",
              "      <td>500</td>\n",
              "      <td>Interventional</td>\n",
              "      <td>Allocation: Randomized. Intervention model: Pa...</td>\n",
              "      <td>Phase 4</td>\n",
              "      <td>NaN</td>\n",
              "      <td></td>\n",
              "      <td>Pinhua Pan, Medical PhD</td>\n",
              "      <td>NaN</td>\n",
              "      <td>pinhuapan668@126.com</td>\n",
              "      <td>+86 13574810968</td>\n",
              "      <td>NaN</td>\n",
              "      <td>\\r&lt;br&gt;        Inclusion Criteria:\\r&lt;br&gt;\\r&lt;br&gt; ...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>2019 Novel Coronavirus</td>\n",
              "      <td>arbidol mg tid conventional group treatment drug</td>\n",
              "      <td>mortality</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>ChiCTR2000029949</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>A Medical Records Based Study for the Effectiv...</td>\n",
              "      <td>A Medical Records Based Study for the Effectiv...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Emergency Department of Zhongnan hospital of W...</td>\n",
              "      <td>2020-02-16</td>\n",
              "      <td>20200216</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ChiCTR</td>\n",
              "      <td>http://www.chictr.org.cn/showproj.aspx?proj=49181</td>\n",
              "      <td>Not Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>18</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Both</td>\n",
              "      <td>2020-02-16</td>\n",
              "      <td>Case series:40;</td>\n",
              "      <td>Observational study</td>\n",
              "      <td>Sequential</td>\n",
              "      <td>Retrospective study</td>\n",
              "      <td>China</td>\n",
              "      <td>Yan, Zhao</td>\n",
              "      <td>NaN</td>\n",
              "      <td>169 Donghu Road, Wuchang District, Wuhan, Hube...</td>\n",
              "      <td>doctoryanzhao@whu.edu.cn</td>\n",
              "      <td>+86 13995577963</td>\n",
              "      <td>Emergency Department of Zhongnan hospital of W...</td>\n",
              "      <td>Inclusion criteria: 1. Aged &gt;=18 years; \\r&lt;br&gt;...</td>\n",
              "      <td>Exclusion criteria: Absolutely forbidden crite...</td>\n",
              "      <td>Novel Coronavirus Pneumonia (COVID-19)</td>\n",
              "      <td>memane case series extracorporeal oxygenation</td>\n",
              "      <td>inhospital length;inhospital mortality;ECMO tr...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>NCT04245631</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>Development of a Simple, Fast and Portable Rec...</td>\n",
              "      <td>Development of a Simple, Fast and Portable Rec...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Beijing Ditan Hospital</td>\n",
              "      <td>26/01/2020</td>\n",
              "      <td>20200126</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ClinicalTrials.gov</td>\n",
              "      <td>https://clinicaltrials.gov/show/NCT04245631</td>\n",
              "      <td>Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>1 Year</td>\n",
              "      <td>90 Years</td>\n",
              "      <td>All</td>\n",
              "      <td>January 1, 2020</td>\n",
              "      <td>50</td>\n",
              "      <td>Observational</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>China</td>\n",
              "      <td>; ;</td>\n",
              "      <td>Yao Xie, Doctor;Yao Xie, Doctor;Yao Xie, Doctor</td>\n",
              "      <td>NaN</td>\n",
              "      <td>;xieyao00120184@sina.com;xieyao00120184@sina.com</td>\n",
              "      <td>;8610-84322200;8610-84322200</td>\n",
              "      <td>Department of Hepatology, Division 2, Beijing ...</td>\n",
              "      <td>\\r&lt;br&gt;        Inclusion Criteria:\\r&lt;br&gt;\\r&lt;br&gt; ...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>New Coronavirus</td>\n",
              "      <td>aided diagnostic amplification assay recombina...</td>\n",
              "      <td>Detection specificity is greater than 95%;Dete...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>NCT04252885</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>The Efficacy of Lopinavir Plus Ritonavir and A...</td>\n",
              "      <td>A Randomized, Open-label, Controlled Study of ...</td>\n",
              "      <td>ELACOI</td>\n",
              "      <td>Guangzhou 8th People's Hospital</td>\n",
              "      <td>30/01/2020</td>\n",
              "      <td>20200130</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ClinicalTrials.gov</td>\n",
              "      <td>https://clinicaltrials.gov/show/NCT04252885</td>\n",
              "      <td>Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>18 Years</td>\n",
              "      <td>80 Years</td>\n",
              "      <td>All</td>\n",
              "      <td>January 28, 2020</td>\n",
              "      <td>125</td>\n",
              "      <td>Interventional</td>\n",
              "      <td>Allocation: Randomized. Intervention model: Pa...</td>\n",
              "      <td>Phase 4</td>\n",
              "      <td>China</td>\n",
              "      <td>;</td>\n",
              "      <td>Yueping Li, MD, Master;Linghua Li, MD, PhD</td>\n",
              "      <td>NaN</td>\n",
              "      <td>lipipi007@163.com;</td>\n",
              "      <td>020- 36052117;</td>\n",
              "      <td>NaN</td>\n",
              "      <td>\\r&lt;br&gt;        Inclusion Criteria:\\r&lt;br&gt;\\r&lt;br&gt; ...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Coronavirus Infections</td>\n",
              "      <td>arbidol lopinavir tablets ritonavir and drug</td>\n",
              "      <td>The rate of virus inhibition</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Yes</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>ChiCTR2000029953</td>\n",
              "      <td>17 February 2020</td>\n",
              "      <td>Construction and Analysis of Prognostic Predic...</td>\n",
              "      <td>Construction and Analysis of Prognostic Predic...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Zhongnan Hospital of Wuhan University</td>\n",
              "      <td>2020-02-17</td>\n",
              "      <td>20200217</td>\n",
              "      <td>4/7/2020 10:40:56 AM</td>\n",
              "      <td>ChiCTR</td>\n",
              "      <td>http://www.chictr.org.cn/showproj.aspx?proj=49217</td>\n",
              "      <td>Not Recruiting</td>\n",
              "      <td>No</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>Both</td>\n",
              "      <td>2020-02-01</td>\n",
              "      <td>survival group:200;died:200;</td>\n",
              "      <td>Observational study</td>\n",
              "      <td>Factorial</td>\n",
              "      <td>NaN</td>\n",
              "      <td>China</td>\n",
              "      <td>Yan Zhao</td>\n",
              "      <td>NaN</td>\n",
              "      <td>169 Donghu Road, Wuchang District, Wuhan, Hube...</td>\n",
              "      <td>doctoryanzhao@whu.edu.cn</td>\n",
              "      <td>+86 13995577963</td>\n",
              "      <td>Emergency Department of Zhongnan Hospital of W...</td>\n",
              "      <td>Inclusion criteria: 2019-nCoV-infected pneumon...</td>\n",
              "      <td>Exclusion criteria: Patients diagnosed with pn...</td>\n",
              "      <td>2019-nCoV Pneumonia</td>\n",
              "      <td>none survival group died</td>\n",
              "      <td>duration of in hospital;in hospital mortality;...</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>NaN</td>\n",
              "      <td>No</td>\n",
              "      <td>False</td>\n",
              "      <td></td>\n",
              "      <td>NaN</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "            TrialID Last Refreshed on  ... results yes no class_label\n",
              "0       NCT04246242   3 February 2020  ...            NaN           5\n",
              "1  ChiCTR2000029949  17 February 2020  ...            NaN           3\n",
              "2       NCT04245631  17 February 2020  ...            NaN           5\n",
              "3       NCT04252885  17 February 2020  ...            NaN           2\n",
              "6  ChiCTR2000029953  17 February 2020  ...            NaN           0\n",
              "\n",
              "[5 rows x 41 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 251
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Cwep9TP4qDuN",
        "colab_type": "text"
      },
      "source": [
        "Next, we will visualize the top 5 keywords for every cluster to analyze the clusters formed"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OvPT9dA95jWq",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def get_top_features_cluster(tf_idf_array, prediction, n_feats):\n",
        "    labels = np.unique(prediction)\n",
        "    dfs = []\n",
        "    for label in labels:\n",
        "        id_temp = np.where(prediction==label) \n",
        "        x_means = np.mean(tf_idf_array[id_temp], axis = 0) \n",
        "        sorted_means = np.argsort(x_means)[::-1][:n_feats] \n",
        "        features = tfidf.get_feature_names()\n",
        "        best_features = [(features[i], x_means[i]) for i in sorted_means]\n",
        "        df_temp = pd.DataFrame(best_features, columns = ['keywords', 'score'])\n",
        "        dfs.append(df_temp)\n",
        "    return dfs\n",
        "dfs = get_top_features_cluster(tf_idf_array, prediction, 5)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "w4Gsctyl6miV",
        "colab_type": "code",
        "outputId": "bfd6c8d5-8ebe-4449-e77a-56f3da0bc719",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "for item in dfs:\n",
        "  ax = item.plot.barh(subplots=True,x='keywords', y='score')\n"
      ],
      "execution_count": 255,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cxvOdsNAqL2a",
        "colab_type": "text"
      },
      "source": [
        "As can be seen, 6 major cluster categories have been formed with these keywords. Some initial insights can be drawn from these clusters. For instance one cluster focuses on drug based interventions, one on testing techniques and standards, one on traditional medicines, another on case series and so on. Medical domain experts might be able to understand and interpret these"
      ]
    }
  ]
}