heartbeat-classification-mit.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "source": [
    "Author: Ani Aggarwal  \n",
    "Github: www.github.com/AniAggarwal"
   ],
   "metadata": {
    "collapsed": false,
    "pycharm": {
     "name": "#%% md\n"
    }
   }
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "unexpected-encounter",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\miniconda3\\envs\\flow\\lib\\site-packages\\ipykernel\\parentpoller.py:113: UserWarning: Parent poll failed.  If the frontend dies,\n",
      "                the kernel may be left running.  Please let us know\n",
      "                about your system (bitness, Python, etc.) at\n",
      "                ipython-dev@scipy.org\n",
      "  warnings.warn(\"\"\"Parent poll failed.  If the frontend dies,\n"
     ]
    }
   ],
   "source": [
    "import librosa\n",
    "import librosa.display\n",
    "import soundfile\n",
    "\n",
    "from scipy.signal import butter, lfilter\n",
    "import numpy as np\n",
    "import sklearn\n",
    "import pandas as pd\n",
    "\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from pathlib import Path\n",
    "import csv\n",
    "import json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "surprising-fiction",
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_SR = 128 # in Hz\n",
    "AUDIO_LEN = 5  # in seconds\n",
    "num_samples = AUDIO_LEN * DATA_SR\n",
    "\n",
    "\n",
    "dataset_dir = Path(\"./datasets/ECG-kaggle-cleaned/\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "apparent-transition",
   "metadata": {},
   "outputs": [],
   "source": [
    "data = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "endangered-contrast",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processing datasets\\ECG-kaggle-cleaned\\incartdb_npy\n",
      "Loading datasets\\ECG-kaggle-cleaned\\incartdb_npy\\I01_BEAT.npy\n",
      "(2757, 2)\n",
      "Processing datasets\\ECG-kaggle-cleaned\\mitdb_npy\n",
      "Loading datasets\\ECG-kaggle-cleaned\\mitdb_npy\\100_BEAT.npy\n",
      "(2273, 2)\n",
      "Processing datasets\\ECG-kaggle-cleaned\\svdb_npy\n",
      "Loading datasets\\ECG-kaggle-cleaned\\svdb_npy\\800_BEAT.npy\n",
      "(1883, 2)\n"
     ]
    }
   ],
   "source": [
    "for subdataset_dir in dataset_dir.iterdir():\n",
    "    if subdataset_dir.is_dir():\n",
    "        print(\"Processing\", subdataset_dir)\n",
    "        with open(subdataset_dir / \"RECORDS\", \"r\") as f:\n",
    "            for line in f:\n",
    "                file_num = line.strip()\n",
    "                data_point = np.load(subdataset_dir / (file_num + \"_BEAT.npy\"))\n",
    "                print(data_point.shape)\n",
    "                \n",
    "                \n",
    "                break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "structured-patrick",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1.25000e+02, 2.57000e+02, 3.86000e+02, ..., 2.30813e+05,\n",
       "       2.30876e+05, 2.31037e+05])"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "array(['N', 'N', 'N', ..., 'V', 'V', 'N'], dtype='<U6')"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test = np.load(\"./datasets/ECG-kaggle-cleaned/mitdb_npy/105_BEAT.npy\", allow_pickle=True)\n",
    "nums = np.array(test[:, 0], dtype=np.float)\n",
    "labels = test[:, 1].astype(str)\n",
    "\n",
    "nums\n",
    "labels"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "id": "fatal-suffering",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<librosa.display.AdaptiveWaveplot at 0x2054e5f7fa0>"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "librosa.display.waveshow(nums, sr=DATA_SR)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "elegant-thesaurus",
   "metadata": {},
   "source": [
    "# Not cleaned dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "excited-consumer",
   "metadata": {},
   "outputs": [],
   "source": [
    "DATA_SR = 125 # in Hz"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "preliminary-fisher",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_train = pd.read_csv(\"./datasets/ECG-kaggle/mitbih_train.csv\", header=None)\n",
    "df_train = df_train.sample(frac=1)\n",
    "df_test = pd.read_csv(\"./datasets/ECG-kaggle/mitbih_test.csv\", header=None)\n",
    "df_test = df_test.sample(frac=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "unlimited-trainer",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = np.array(df_train[list(range(187))].values)\n",
    "X_train = np.expand_dims(X_train, -1)\n",
    "y_train = np.array(df_train[187].values).astype(np.int8)\n",
    "y_train[y_train != 0] = 1 # coverting to be just normal (0) and abnormal (1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "proud-brass",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_test = np.array(df_test[list(range(187))].values)\n",
    "X_test = np.expand_dims(X_test, -1)\n",
    "y_test = np.array(df_test[187].values).astype(np.int8)\n",
    "y_test[y_test != 0] = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "consistent-filing",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0, 1}"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "(87554, 187, 1)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "(87554,)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "(21892, 187, 1)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "text/plain": [
       "(21892,)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "set(y_train)\n",
    "\n",
    "X_train.shape\n",
    "y_train.shape\n",
    "X_test.shape\n",
    "y_test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "finite-compiler",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(187, 1)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train[0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "revised-public",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(187,)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train[0, :, 0].shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "egyptian-officer",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1.0, 0.0, 0.17428295254906395)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train.max(), X_train.min(), X_train.mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "expensive-parker",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Anaconda3\\miniconda3\\envs\\flow\\lib\\site-packages\\librosa\\util\\utils.py:198: UserWarning: librosa.util.frame called with axis=-1 on a non-contiguous input. This will result in a copy.\n",
      "  warnings.warn(\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<librosa.display.AdaptiveWaveplot at 0x17db631a460>"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "librosa.display.waveshow(X_train[0, :, 0], sr=DATA_SR)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "editorial-lewis",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "soundfile.write(\"./datasets/test-wav-normal.wav\", X_train[0, :, 0], 125)\n",
    "y_train[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "genetic-eating",
   "metadata": {},
   "source": [
    "# Inference"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "editorial-occurrence",
   "metadata": {},
   "outputs": [],
   "source": [
    "# model_full_save = Path(\n",
    "#     \"./model_saves/transfer/2021-06-27_05-35-15_epochs_100-batch_size_1000-lr_0.01/full_save\"\n",
    "# )\n",
    "model_full_save = Path(\n",
    "    \"./model_saves/best-0.9889_val_acc-backup/full_save\"\n",
    ")\n",
    "\n",
    "test_audio_path = Path(\n",
    "    \"./datasets/classification-heart-sounds-physionet/training-a/a0004.wav\"\n",
    ")\n",
    "test_set_path = Path(\n",
    "    \"./datasets/classification-heart-sounds-physionet/validation\"\n",
    ")\n",
    "test_output_path = Path(\"./datasets/classification-heart-sounds-physionet/numpy-data/data-test.json\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "greatest-conversion",
   "metadata": {},
   "outputs": [],
   "source": [
    "from inference import Inference\n",
    "from create_dataset import create_test_set\n",
    "from clean_dataset import load_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "cosmetic-juvenile",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "bidirectional (Bidirectional (None, 187, 256)          133120    \n",
      "_________________________________________________________________\n",
      "bidirectional_1 (Bidirection (None, 187, 128)          164352    \n",
      "_________________________________________________________________\n",
      "bidirectional_2 (Bidirection (None, 64)                41216     \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 32)                2080      \n",
      "_________________________________________________________________\n",
      "batch_normalization (BatchNo (None, 32)                128       \n",
      "_________________________________________________________________\n",
      "re_lu (ReLU)                 (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 1)                 33        \n",
      "=================================================================\n",
      "Total params: 340,929\n",
      "Trainable params: 340,865\n",
      "Non-trainable params: 64\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "inference = Inference(\n",
    "    model_full_save, 125, 125\n",
    ")  # for testing, dataset has 2kHz sample rate"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "editorial-breeding",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[('unsure', array([0], dtype=int8)),\n",
       " ('unsure', array([1], dtype=int8)),\n",
       " ('unsure', array([0], dtype=int8)),\n",
       " ('unsure', array([0], dtype=int8)),\n",
       " ('unsure', array([0], dtype=int8)),\n",
       " ('unsure', array([1], dtype=int8)),\n",
       " ('unsure', array([1], dtype=int8)),\n",
       " ('unsure', array([0], dtype=int8)),\n",
       " ('unsure', array([0], dtype=int8)),\n",
       " ('unsure', array([0], dtype=int8))]"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[(inference.predict(X_train[i:i+1]), y_train[i:i+1]) for i in range(10)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "infrared-charger",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Data loaded. X shape: (4093, 187), y shape: (4093,)\n"
     ]
    }
   ],
   "source": [
    "# create_test_set(test_set_path, test_output_path)\n",
    "X_test, y_test = load_data(test_output_path)\n",
    "X_test = np.expand_dims(X_test, -1)  # add dimension to make it uniform with model input\n",
    "y_test[y_test == -1] = 0  # convert to 1 for abnormal and 0 for normal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "outdoor-tutorial",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "128/128 [==============================] - 3s 23ms/step - loss: 1.2352 - accuracy: 0.0000e+00 - binary_accuracy: 0.4977 - AUC-ROC: 0.4955 - AUC-PR: 0.5107 - precision: 0.5026 - recall: 0.6498 - true_positive: 1345.0000 - false_positive: 1331.0000 - false_negative: 725.0000 - true_negative: 692.0000\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "{'loss': 1.2352006435394287,\n",
       " 'accuracy': 0.0,\n",
       " 'binary_accuracy': 0.4976789653301239,\n",
       " 'AUC-ROC': 0.49552395939826965,\n",
       " 'AUC-PR': 0.5106804370880127,\n",
       " 'precision': 0.5026158690452576,\n",
       " 'recall': 0.6497584581375122,\n",
       " 'true_positive': 1345.0,\n",
       " 'false_positive': 1331.0,\n",
       " 'false_negative': 725.0,\n",
       " 'true_negative': 692.0,\n",
       " 'f1': 0.5667931060295063,\n",
       " 'sensitivity': 0.6497584541062802,\n",
       " 'specificity': 0.34206623826000987}"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result = inference.evaluate(X_test, y_test)\n",
    "\n",
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "graduate-filter",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tmp\\tmp_recording.wav\n"
     ]
    }
   ],
   "source": [
    "audio_np, sr = librosa.load(test_audio_path, sr=125)\n",
    "wav = inference.get_wav(audio_np.tolist())\n",
    "print(wav)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "refined-hello",
   "metadata": {},
   "outputs": [],
   "source": [
    "inference.delete_wav(wav)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "connected-westminster",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "abnormal\n"
     ]
    }
   ],
   "source": [
    "print(inference.predict(audio_np))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "cheap-netscape",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:flow]",
   "language": "python",
   "name": "conda-env-flow-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}