unet_for_ecg.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Provando ad implementare il codice di un articolo\n",
    "\n",
    "- [Ispirazione](https://paperswithcode.com/paper/deep-learning-for-ecg-segmentation): PapersWithCode\n",
    "- [Paper](https://arxiv.org/pdf/2001.04689.pdf): Deep Learning for ECG Segmentation\n",
    "- [Dataset](https://physionet.org/content/ludb/1.0.1): Lobachevsky University Electrocardiography Database\n",
    "- [Altro Notebook](https://www.kaggle.com/yschii/find-ecg-wave): My Kaggle\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already up-to-date: wfdb in c:\\users\\scala\\lang\\lib\\site-packages (3.4.1)\n",
      "Requirement already satisfied, skipping upgrade: requests>=2.8.1 in c:\\users\\scala\\lang\\lib\\site-packages (from wfdb) (2.25.1)Note: you may need to restart the kernel to use updated packages.\n",
      "Requirement already satisfied, skipping upgrade: pandas>=0.17.0 in c:\\users\\scala\\lang\\lib\\site-packages (from wfdb) (1.3.5)\n",
      "Requirement already satisfied, skipping upgrade: matplotlib>=3.3.4 in c:\\users\\scala\\lang\\lib\\site-packages (from wfdb) (3.5.1)\n",
      "Requirement already satisfied, skipping upgrade: numpy>=1.10.1 in c:\\users\\scala\\lang\\lib\\site-packages (from wfdb) (1.18.5)\n",
      "Requirement already satisfied, skipping upgrade: scipy>=0.17.0 in c:\\users\\scala\\lang\\lib\\site-packages (from wfdb) (1.5.2)\n",
      "Requirement already satisfied, skipping upgrade: certifi>=2017.4.17 in c:\\users\\scala\\lang\\lib\\site-packages (from requests>=2.8.1->wfdb) (2019.11.28)\n",
      "Requirement already satisfied, skipping upgrade: idna<3,>=2.5 in c:\\users\\scala\\lang\\lib\\site-packages (from requests>=2.8.1->wfdb) (2.9)\n",
      "Requirement already satisfied, skipping upgrade: urllib3<1.27,>=1.21.1 in c:\\users\\scala\\lang\\lib\\site-packages (from requests>=2.8.1->wfdb) (1.25.8)\n",
      "Requirement already satisfied, skipping upgrade: chardet<5,>=3.0.2 in c:\\users\\scala\\lang\\lib\\site-packages (from requests>=2.8.1->wfdb) (3.0.4)\n",
      "Requirement already satisfied, skipping upgrade: pytz>=2017.3 in c:\\users\\scala\\lang\\lib\\site-packages (from pandas>=0.17.0->wfdb) (2019.3)\n",
      "Requirement already satisfied, skipping upgrade: python-dateutil>=2.7.3 in c:\\users\\scala\\lang\\lib\\site-packages (from pandas>=0.17.0->wfdb) (2.8.1)\n",
      "Requirement already satisfied, skipping upgrade: kiwisolver>=1.0.1 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib>=3.3.4->wfdb) (1.3.1)\n",
      "Requirement already satisfied, skipping upgrade: pillow>=6.2.0 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib>=3.3.4->wfdb) (7.1.2)\n",
      "Requirement already satisfied, skipping upgrade: pyparsing>=2.2.1 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib>=3.3.4->wfdb) (2.4.7)\n",
      "Requirement already satisfied, skipping upgrade: cycler>=0.10 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib>=3.3.4->wfdb) (0.10.0)\n",
      "Requirement already satisfied, skipping upgrade: packaging>=20.0 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib>=3.3.4->wfdb) (20.8)\n",
      "Requirement already satisfied, skipping upgrade: fonttools>=4.22.0 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib>=3.3.4->wfdb) (4.28.5)\n",
      "Requirement already satisfied, skipping upgrade: six>=1.5 in c:\\users\\scala\\lang\\lib\\site-packages (from python-dateutil>=2.7.3->pandas>=0.17.0->wfdb) (1.14.0)\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: You are using pip version 20.0.2; however, version 21.3.1 is available.\n",
      "You should consider upgrading via the 'C:\\Users\\scala\\lang\\python.exe -m pip install --upgrade pip' command.\n",
      "WARNING: You are using pip version 20.0.2; however, version 21.3.1 is available.\n",
      "You should consider upgrading via the 'C:\\Users\\scala\\lang\\python.exe -m pip install --upgrade pip' command.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already up-to-date: scikit-learn in c:\\users\\scala\\lang\\lib\\site-packages (1.0.2)\n",
      "Requirement already up-to-date: tensorflow in c:\\users\\scala\\lang\\lib\\site-packages (2.7.0)\n",
      "Requirement already up-to-date: matplotlib in c:\\users\\scala\\lang\\lib\\site-packages (3.5.1)\n",
      "Requirement already up-to-date: pandas in c:\\users\\scala\\lang\\lib\\site-packages (1.3.5)\n",
      "Requirement already satisfied, skipping upgrade: scipy>=1.1.0 in c:\\users\\scala\\lang\\lib\\site-packages (from scikit-learn) (1.5.2)\n",
      "Requirement already satisfied, skipping upgrade: threadpoolctl>=2.0.0 in c:\\users\\scala\\lang\\lib\\site-packages (from scikit-learn) (2.1.0)\n",
      "Requirement already satisfied, skipping upgrade: numpy>=1.14.6 in c:\\users\\scala\\lang\\lib\\site-packages (from scikit-learn) (1.18.5)\n",
      "Requirement already satisfied, skipping upgrade: joblib>=0.11 in c:\\users\\scala\\lang\\lib\\site-packages (from scikit-learn) (0.17.0)\n",
      "Requirement already satisfied, skipping upgrade: grpcio<2.0,>=1.24.3 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (1.43.0)\n",
      "Requirement already satisfied, skipping upgrade: google-pasta>=0.1.1 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (0.2.0)\n",
      "Requirement already satisfied, skipping upgrade: astunparse>=1.6.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (1.6.3)\n",
      "Requirement already satisfied, skipping upgrade: absl-py>=0.4.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (1.0.0)\n",
      "Requirement already satisfied, skipping upgrade: libclang>=9.0.1 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (12.0.0)\n",
      "Requirement already satisfied, skipping upgrade: gast<0.5.0,>=0.2.1 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (0.4.0)\n",
      "Requirement already satisfied, skipping upgrade: tensorflow-estimator<2.8,~=2.7.0rc0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (2.7.0)\n",
      "Requirement already satisfied, skipping upgrade: typing-extensions>=3.6.6 in c:\\users\\scala\\appdata\\roaming\\python\\python38\\site-packages (from tensorflow) (3.7.4.3)\n",
      "Requirement already satisfied, skipping upgrade: keras<2.8,>=2.7.0rc0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (2.7.0)\n",
      "Requirement already satisfied, skipping upgrade: h5py>=2.9.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (3.6.0)\n",
      "Requirement already satisfied, skipping upgrade: opt-einsum>=2.3.2 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (3.3.0)\n",
      "Requirement already satisfied, skipping upgrade: flatbuffers<3.0,>=1.12 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (2.0)\n",
      "Requirement already satisfied, skipping upgrade: tensorflow-io-gcs-filesystem>=0.21.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (0.23.1)\n",
      "Requirement already satisfied, skipping upgrade: protobuf>=3.9.2 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (3.12.2)\n",
      "Requirement already satisfied, skipping upgrade: six>=1.12.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (1.14.0)\n",
      "Requirement already satisfied, skipping upgrade: termcolor>=1.1.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (1.1.0)\n",
      "Requirement already satisfied, skipping upgrade: wheel<1.0,>=0.32.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (0.37.1)\n",
      "Requirement already satisfied, skipping upgrade: wrapt>=1.11.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (1.13.3)\n",
      "Requirement already satisfied, skipping upgrade: keras-preprocessing>=1.1.1 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (1.1.2)\n",
      "Requirement already satisfied, skipping upgrade: tensorboard~=2.6 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorflow) (2.7.0)\n",
      "Requirement already satisfied, skipping upgrade: python-dateutil>=2.7 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib) (2.8.1)\n",
      "Requirement already satisfied, skipping upgrade: cycler>=0.10 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib) (0.10.0)\n",
      "Requirement already satisfied, skipping upgrade: kiwisolver>=1.0.1 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib) (1.3.1)\n",
      "Requirement already satisfied, skipping upgrade: pillow>=6.2.0 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib) (7.1.2)Note: you may need to restart the kernel to use updated packages.\n",
      "Requirement already satisfied, skipping upgrade: pyparsing>=2.2.1 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib) (2.4.7)\n",
      "Requirement already satisfied, skipping upgrade: fonttools>=4.22.0 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib) (4.28.5)\n",
      "Requirement already satisfied, skipping upgrade: packaging>=20.0 in c:\\users\\scala\\lang\\lib\\site-packages (from matplotlib) (20.8)\n",
      "Requirement already satisfied, skipping upgrade: pytz>=2017.3 in c:\\users\\scala\\lang\\lib\\site-packages (from pandas) (2019.3)\n",
      "Requirement already satisfied, skipping upgrade: setuptools in c:\\users\\scala\\lang\\lib\\site-packages (from protobuf>=3.9.2->tensorflow) (41.2.0)\n",
      "Requirement already satisfied, skipping upgrade: google-auth<3,>=1.6.3 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorboard~=2.6->tensorflow) (2.3.3)\n",
      "Requirement already satisfied, skipping upgrade: tensorboard-plugin-wit>=1.6.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorboard~=2.6->tensorflow) (1.8.0)\n",
      "Requirement already satisfied, skipping upgrade: tensorboard-data-server<0.7.0,>=0.6.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorboard~=2.6->tensorflow) (0.6.1)\n",
      "Requirement already satisfied, skipping upgrade: google-auth-oauthlib<0.5,>=0.4.1 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorboard~=2.6->tensorflow) (0.4.6)\n",
      "Requirement already satisfied, skipping upgrade: requests<3,>=2.21.0 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorboard~=2.6->tensorflow) (2.25.1)\n",
      "Requirement already satisfied, skipping upgrade: werkzeug>=0.11.15 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorboard~=2.6->tensorflow) (1.0.0)\n",
      "Requirement already satisfied, skipping upgrade: markdown>=2.6.8 in c:\\users\\scala\\lang\\lib\\site-packages (from tensorboard~=2.6->tensorflow) (3.3.6)\n",
      "Requirement already satisfied, skipping upgrade: pyasn1-modules>=0.2.1 in c:\\users\\scala\\lang\\lib\\site-packages (from google-auth<3,>=1.6.3->tensorboard~=2.6->tensorflow) (0.2.8)\n",
      "Requirement already satisfied, skipping upgrade: cachetools<5.0,>=2.0.0 in c:\\users\\scala\\lang\\lib\\site-packages (from google-auth<3,>=1.6.3->tensorboard~=2.6->tensorflow) (4.2.4)\n",
      "Requirement already satisfied, skipping upgrade: rsa<5,>=3.1.4; python_version >= \"3.6\" in c:\\users\\scala\\lang\\lib\\site-packages (from google-auth<3,>=1.6.3->tensorboard~=2.6->tensorflow) (4.0)\n",
      "Requirement already satisfied, skipping upgrade: requests-oauthlib>=0.7.0 in c:\\users\\scala\\lang\\lib\\site-packages (from google-auth-oauthlib<0.5,>=0.4.1->tensorboard~=2.6->tensorflow) (1.3.0)\n",
      "Requirement already satisfied, skipping upgrade: certifi>=2017.4.17 in c:\\users\\scala\\lang\\lib\\site-packages (from requests<3,>=2.21.0->tensorboard~=2.6->tensorflow) (2019.11.28)\n",
      "Requirement already satisfied, skipping upgrade: urllib3<1.27,>=1.21.1 in c:\\users\\scala\\lang\\lib\\site-packages (from requests<3,>=2.21.0->tensorboard~=2.6->tensorflow) (1.25.8)\n",
      "Requirement already satisfied, skipping upgrade: idna<3,>=2.5 in c:\\users\\scala\\lang\\lib\\site-packages (from requests<3,>=2.21.0->tensorboard~=2.6->tensorflow) (2.9)\n",
      "Requirement already satisfied, skipping upgrade: chardet<5,>=3.0.2 in c:\\users\\scala\\lang\\lib\\site-packages (from requests<3,>=2.21.0->tensorboard~=2.6->tensorflow) (3.0.4)\n",
      "Requirement already satisfied, skipping upgrade: importlib-metadata>=4.4; python_version < \"3.10\" in c:\\users\\scala\\lang\\lib\\site-packages (from markdown>=2.6.8->tensorboard~=2.6->tensorflow) (4.10.0)\n",
      "Requirement already satisfied, skipping upgrade: pyasn1<0.5.0,>=0.4.6 in c:\\users\\scala\\lang\\lib\\site-packages (from pyasn1-modules>=0.2.1->google-auth<3,>=1.6.3->tensorboard~=2.6->tensorflow) (0.4.8)\n",
      "\n",
      "Requirement already satisfied, skipping upgrade: oauthlib>=3.0.0 in c:\\users\\scala\\lang\\lib\\site-packages (from requests-oauthlib>=0.7.0->google-auth-oauthlib<0.5,>=0.4.1->tensorboard~=2.6->tensorflow) (3.1.1)\n",
      "Requirement already satisfied, skipping upgrade: zipp>=0.5 in c:\\users\\scala\\lang\\lib\\site-packages (from importlib-metadata>=4.4; python_version < \"3.10\"->markdown>=2.6.8->tensorboard~=2.6->tensorflow) (3.6.0)\n"
     ]
    }
   ],
   "source": [
    "# first of all install necessary packages\n",
    "# wfdb usefull for reading ECG data\n",
    "%pip install -U wfdb\n",
    "# usefull for machine learning\n",
    "%pip install -U scikit-learn tensorflow matplotlib pandas\n",
    "# set matplotlib to plot in the notebook\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# then import necessary packages\n",
    "import wfdb\n",
    "from os.path import exists as file_exists\n",
    "import numpy as np \n",
    "import pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import tensorflow as tf\n",
    "import tensorflow.keras.backend as K\n",
    "from tensorflow.keras.models import load_model\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.layers import Input, Conv1DTranspose, BatchNormalization, Activation, Conv1D, ZeroPadding1D, MaxPooling1D, Concatenate\n",
    "from tensorflow.keras.optimizers import Adam, SGD, RMSprop\n",
    "from tensorflow.keras.metrics import MeanIoU\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "\n",
    "tf.config.run_functions_eagerly(False)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Explore some data\n",
    "\n",
    "As described in the paper, each ecg has 12 leads, is long 10seconds and since it s sampled at 500hz it has 5000 samples\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fisrt set some constants usefull along the code\n",
    "\n",
    "There 4 constants\n",
    " - leads of the ecg\n",
    " - location of the data\n",
    " - some colors for plotting\n",
    " - the size of the ECG the model is fed with"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# since the paper handles 12 lead ECG data, i will use 12 leads\n",
    "LEADS = ['avf', 'avl', 'avr', 'i', 'ii', 'iii', 'v1', 'v2', 'v3', 'v4', 'v5', 'v6']\n",
    "DATA_FOLDER = \"./lobachevsky-university-electrocardiography-database-1.0.1/data\"\n",
    "SEGMENT_TO_COLOR = {\n",
    "    'p': 'red',\n",
    "    'N': 'blue',\n",
    "    't': 'green',\n",
    "}\n",
    "ECG_SIZE = 41*16 # -> 32*16 = 512\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then some function useful to plot and manipulate the ECGs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "from typing import Union,List, Tuple\n",
    "\n",
    "# get signal with wfdb library\n",
    "# using an index\n",
    "def get_signal(index:int, as_p_signal:bool=True) -> Union[wfdb.Record, np.ndarray]:\n",
    "    record = wfdb.rdrecord(DATA_FOLDER + \"/\" +  str(index))\n",
    "    assert type(record) is wfdb.Record\n",
    "\n",
    "    if as_p_signal:\n",
    "        assert type(record.p_signal) is np.ndarray\n",
    "        return record.p_signal\n",
    "        \n",
    "    return record\n",
    "\n",
    "# get annotations given the ecg lead\n",
    "def get_annotations(index:int, lead, as_sample=True) -> Union[wfdb.Annotation, np.ndarray]:\n",
    "    annotations = wfdb.rdann(DATA_FOLDER +\"/\" + str(index), extension = lead)\n",
    "    if as_sample:\n",
    "        return np.array(annotations.sample)\n",
    "    return annotations\n",
    "\n",
    "# get a full EGC with 12 leads\n",
    "def get_full_ecg(index:int):\n",
    "    signal = get_signal(index)\n",
    "    annotations = [\n",
    "        get_annotations(index, lead) for lead in LEADS\n",
    "    ]\n",
    "    return signal, annotations\n",
    "\n",
    "\n",
    "def get_single_lead_ecg(index, lead) -> Tuple[np.ndarray, np.ndarray]:\n",
    "    \"\"\"\n",
    "    return and ecg signal and its annotations\n",
    "    both as ndarray\n",
    "    \"\"\"\n",
    "    signal = get_signal(index)\n",
    "    assert type(signal) is np.ndarray\n",
    "    signal = signal[:, LEADS.index(lead)]\n",
    "\n",
    "    samples = get_annotations(index, lead)\n",
    "    assert type(samples) is np.ndarray\n",
    "    \n",
    "    return signal, samples\n",
    "\n",
    "\n",
    "def get_annotations_symbols(index, lead):\n",
    "    ann = get_annotations(index, lead, as_sample=False)\n",
    "    return ann.symbol\n",
    "\n",
    "def paired_annotation_sample_and_symbol(index, lead):\n",
    "    annotations_symbols = get_annotations_symbols(index, lead)\n",
    "    annotations_sample = get_annotations(index, lead)\n",
    "    return zip(annotations_sample, annotations_symbols)\n",
    "\n",
    "\n",
    "def get_single_lead_ecg_with_symbols(index, lead):\n",
    "    \"\"\"\n",
    "    return and ecg signal and its annotations\n",
    "    both as ndarray\n",
    "    \"\"\"\n",
    "    signal = get_signal(index)\n",
    "    assert type(signal) is np.ndarray\n",
    "    signal = signal[:, LEADS.index(lead)]\n",
    "\n",
    "    data = paired_annotation_sample_and_symbol(index, lead)\n",
    "    \n",
    "    return signal, np.array(list(data))\n",
    "\n",
    "# plot single lead ecg with annotations\n",
    "def plot_single_lead_ecg(index, lead):\n",
    "    signal, samples = get_single_lead_ecg(index, lead)\n",
    "\n",
    "    fig, ax = plt.subplots(figsize = (28, 3))\n",
    "\n",
    "    ax.plot(signal)\n",
    "    ax.scatter(samples, signal[samples], c = 'r', marker = 'o')\n",
    "\n",
    "\n",
    "# now plot every lead with annotations\n",
    "def plot_signal_with_annotation(index):\n",
    "    signal, samples = get_full_ecg(index)\n",
    "    # extract sample from annotations\n",
    "    wfdb.plot_items(signal, samples )\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "Every lead is divided in 3 parts\n",
    " - the signal `[0.32, 0.454, 0.112, 0.461, 0.28]`\n",
    " - the annotation `['(', 'p', ')', '(']`\n",
    " - the time points at wich annotation are registered `[2, 5, 11, 15]` like index for the signal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 12 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_signal_with_annotation(index = 199)   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Still, the ECGs are a bit hard to handle, so i write a class to handle them."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "Generator that allows to iterate over a list, grouping the elements in groups of n.\n",
    "Example:\n",
    "    >>> list(grouped([1,2,3,4,5,6,7,8,9], n=3))\n",
    "    [(1, 2, 3), (4, 5, 6), (7, 8, 9)]\n",
    "\"\"\"\n",
    "def grouped(itr, n=3):\n",
    "    itr = iter(itr)\n",
    "    end = object()\n",
    "    while True:\n",
    "        vals = tuple(next(itr, end) for _ in range(n))\n",
    "        if vals[-1] is end:\n",
    "            return\n",
    "        yield vals"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2016x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "class EGCSignal:\n",
    "    \"\"\"\n",
    "    This class has 4 main purposes:\n",
    "    1. To store the signal and its annotations\n",
    "    2. To cut the signal once at the beginning and once at the end\n",
    "    3. To plot the ECG in different ways\n",
    "    4. To convert the annotation in a one hot encoding\n",
    "\n",
    "    Note that doesn't store the entire ECG, but only one lead\n",
    "\n",
    "    Also has a method to initialize the class without explicitly passing the signal and annotations\n",
    "    but with the index and lead of the record\n",
    "    \"\"\"\n",
    "    def __init__(self, signal, time_points, symbol, categories = None):\n",
    "        self.signal:np.ndarray = signal\n",
    "        self.time_points:np.ndarray = time_points\n",
    "        self.symbols:list[str] = symbol\n",
    "        self.symbol_to_category = {\n",
    "            'N': 0,\n",
    "            't': 1,\n",
    "            'p': 2\n",
    "        }\n",
    "        self.category_to_symbol = {\n",
    "            0: 'N',\n",
    "            1: 't',\n",
    "            2: 'p'\n",
    "        }\n",
    "        self.categories = categories if categories is not None else self.symbols_to_category()\n",
    "        self._cut_beginning( 550)\n",
    "        self._cut_end( 3500)\n",
    "\n",
    "    def __getitem__(self, key):\n",
    "        return self.signal[key]\n",
    "\n",
    "    def __len__(self):\n",
    "        return len(self.signal)\n",
    "\n",
    "    def _cut_beginning(self, start_point):\n",
    "        self.signal = self.signal[start_point:]\n",
    "        self.categories = self.categories[start_point:]\n",
    "\n",
    "        # now have to check if time_points and symbols are also to cut\n",
    "        if start_point > self.time_points[0]:\n",
    "            # get the index of the first time point greater than start_point\n",
    "            index = np.argmax(self.time_points > start_point)                \n",
    "            self.time_points = self.time_points[index:]\n",
    "            self.symbols = self.symbols[index:]\n",
    "\n",
    "        self.time_points = self.time_points - start_point\n",
    "\n",
    "        # check the cut point\n",
    "        if self.categories[0] != -1:\n",
    "            # if the first symbol is a ')' then i have to prepend a '(' and a letter from self.category_to_symbol\n",
    "            if self.symbols[0] == ')':\n",
    "                self.symbols = ['('] + [self.category_to_symbol[self.categories[0]]] + self.symbols\n",
    "                self.time_points = np.concatenate(([0,1], self.time_points))\n",
    "            elif self.symbols[0] in self.symbol_to_category:\n",
    "                # just prepend '('\n",
    "                self.symbols = ['('] + self.symbols\n",
    "                self.time_points = np.concatenate(([0], self.time_points))\n",
    "\n",
    "    def _cut_end(self, end_point):\n",
    "        self.signal = self.signal[:end_point]\n",
    "        self.categories = self.categories[:end_point]\n",
    "\n",
    "        index = self.time_points[self.time_points < self.signal.size].size\n",
    "        self.time_points = self.time_points[:index]\n",
    "        self.symbols = self.symbols[:index]\n",
    "\n",
    "        # check the cut point\n",
    "        if self.categories[-1] != -1:\n",
    "            # if the last symbol is a '(' then i have to append a ')' and a letter from self.category_to_symbol\n",
    "            if self.symbols[-1] == '(':\n",
    "                self.symbols = self.symbols +  [self.category_to_symbol[self.categories[-1]]] + [')']\n",
    "                self.time_points = np.concatenate((self.time_points, [self.signal.size - 1, self.signal.size ]))\n",
    "            elif self.symbols[-1] in self.symbol_to_category:\n",
    "                # just append ')'\n",
    "                self.symbols = self.symbols + [')']\n",
    "                self.time_points = np.concatenate((self.time_points, [self.signal.size]))\n",
    " \n",
    "    def plot(self, ax=None):\n",
    "        if ax is None:\n",
    "            fig, ax = plt.subplots(figsize = (28, 3))\n",
    "        ax.plot(self.signal)\n",
    "\n",
    "    def plot_with_time_point(self):\n",
    "        fig, ax = plt.subplots(figsize = (28, 3))\n",
    "        self.plot(ax)\n",
    "        ax.scatter(self.time_points, self.signal[self.time_points], c = 'r', marker = 'o')\n",
    "\n",
    "    def plot_with_segments(self):\n",
    "        fig, ax = plt.subplots(figsize = (28, 3))\n",
    "        self.plot(ax)\n",
    "\n",
    "        for start, symbol, end in grouped(self.time_points, 3):            \n",
    "            i = np.nonzero(self.time_points == symbol)[0][0]\n",
    "            current_symbol = self.symbols[i]\n",
    "            color = SEGMENT_TO_COLOR[current_symbol]\n",
    "            ax.axvspan(start, end, color=color, alpha=0.4)\n",
    "\n",
    "    def symbols_to_category(self):\n",
    "        \"\"\"\n",
    "        converts the symbols list in a numpy array of integers\n",
    "        same length as the signal\n",
    "        \"\"\"\n",
    "\n",
    "        # first instantiate an array of -1 same length as the signal\n",
    "        category = np.full(len(self.signal), -1)\n",
    "        # now fill the array with the known category\n",
    "        for section in grouped(self.time_points):\n",
    "            # unpack the section\n",
    "            start, peak, end = section\n",
    "            \n",
    "            \n",
    "            # get the category given the peak\n",
    "            i = np.nonzero(self.time_points == peak)[0][0]\n",
    "            current_symbol = self.symbols[i]\n",
    "\n",
    "            category[start:end] = self.symbol_to_category[current_symbol]\n",
    "\n",
    "        return category\n",
    "\n",
    "    @staticmethod\n",
    "    def from_index_and_lead(index, lead):\n",
    "        return EGCSignal(\n",
    "        get_signal(index)[:, LEADS.index(lead)],\n",
    "        get_annotations(index, lead),\n",
    "        get_annotations_symbols(index, lead) )\n",
    "\n",
    "\n",
    "\n",
    "EGCSignal.from_index_and_lead(1, LEADS[0]).plot_with_segments()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Dateset preprocessing\n",
    "\n",
    "Now i use the ECG class to cut parts of the dataset and create the train/test set\n",
    "\n",
    "For some reasons, a few hundreds of signals are not correctly loaded.\n",
    "\n",
    "The full dataset will be:\n",
    "```\n",
    "200 * 12 * 4 = 9600\n",
    "9600 - ~200 => 9400 (still a large dataset, i will not dig in to why some ECGs are not loaded)\n",
    "```\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, "
     ]
    }
   ],
   "source": [
    "import random\n",
    "import copy\n",
    "\n",
    "\n",
    "X = []\n",
    "Y = []\n",
    "for ecg_index in range(1,200):\n",
    "    print(ecg_index, end = ', ')\n",
    "    for lead in LEADS:\n",
    "        try:\n",
    "            start = 0\n",
    "            s = EGCSignal.from_index_and_lead(ecg_index, lead)\n",
    "            for ecg_part in range(0, 4):\n",
    "                s_copy = copy.deepcopy(s)\n",
    "\n",
    "                span = ECG_SIZE\n",
    "                start += random.randint(0, 30) + span\n",
    "                s_copy._cut_beginning(start)\n",
    "                s_copy._cut_end(span)\n",
    "\n",
    "                # append the signal to X, also adding a new dimension\n",
    "                X.append(s_copy.signal)\n",
    "                # append the category to Y\n",
    "                Y.append(to_categorical(s_copy.categories + 1, num_classes = 4).astype(np.float32))\n",
    "        except:\n",
    "            pass\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model Creation and Train\n",
    "\n",
    "The model is described as a 1-dimentional UNet"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model_1\"\n",
      "__________________________________________________________________________________________________\n",
      " Layer (type)                   Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      " input_2 (InputLayer)           [(None, 656, 1)]     0           []                               \n",
      "                                                                                                  \n",
      " conv_pass_layer_30 (ConvPassLa  (None, 656, 4)      56          ['input_2[0][0]']                \n",
      " yer)                                                                                             \n",
      "                                                                                                  \n",
      " conv_pass_layer_31 (ConvPassLa  (None, 656, 4)      164         ['conv_pass_layer_30[0][0]']     \n",
      " yer)                                                                                             \n",
      "                                                                                                  \n",
      " down_pass_layer_3 (DownPassLay  (None, 328, 8)      944         ['conv_pass_layer_31[0][0]']     \n",
      " er)                                                                                              \n",
      "                                                                                                  \n",
      " down_pass_layer_4 (DownPassLay  (None, 164, 16)     3616        ['down_pass_layer_3[0][0]']      \n",
      " er)                                                                                              \n",
      "                                                                                                  \n",
      " down_pass_layer_5 (DownPassLay  (None, 82, 32)      14144       ['down_pass_layer_4[0][0]']      \n",
      " er)                                                                                              \n",
      "                                                                                                  \n",
      " max_pooling1d_10 (MaxPooling1D  (None, 41, 32)      0           ['down_pass_layer_5[0][0]']      \n",
      " )                                                                                                \n",
      "                                                                                                  \n",
      " conv_pass_layer_38 (ConvPassLa  (None, 41, 64)      18752       ['max_pooling1d_10[0][0]']       \n",
      " yer)                                                                                             \n",
      "                                                                                                  \n",
      " conv_pass_layer_39 (ConvPassLa  (None, 41, 64)      37184       ['conv_pass_layer_38[0][0]']     \n",
      " yer)                                                                                             \n",
      "                                                                                                  \n",
      " up_pass_layer_3 (UpPassLayer)  (None, 82, 32)       70016       ['down_pass_layer_5[0][0]',      \n",
      "                                                                  'conv_pass_layer_39[0][0]']     \n",
      "                                                                                                  \n",
      " up_pass_layer_4 (UpPassLayer)  (None, 164, 16)      17600       ['down_pass_layer_4[0][0]',      \n",
      "                                                                  'up_pass_layer_3[0][0]']        \n",
      "                                                                                                  \n",
      " up_pass_layer_5 (UpPassLayer)  (None, 328, 8)       4448        ['down_pass_layer_3[0][0]',      \n",
      "                                                                  'up_pass_layer_4[0][0]']        \n",
      "                                                                                                  \n",
      " conv1d_transpose_10 (Conv1DTra  (None, 656, 8)      520         ['up_pass_layer_5[0][0]']        \n",
      " nspose)                                                                                          \n",
      "                                                                                                  \n",
      " concatenate_10 (Concatenate)   (None, 656, 12)      0           ['conv1d_transpose_10[0][0]',    \n",
      "                                                                  'conv_pass_layer_31[0][0]']     \n",
      "                                                                                                  \n",
      " conv_pass_layer_46 (ConvPassLa  (None, 656, 4)      452         ['concatenate_10[0][0]']         \n",
      " yer)                                                                                             \n",
      "                                                                                                  \n",
      " conv_pass_layer_47 (ConvPassLa  (None, 656, 4)      164         ['conv_pass_layer_46[0][0]']     \n",
      " yer)                                                                                             \n",
      "                                                                                                  \n",
      "==================================================================================================\n",
      "Total params: 168,060\n",
      "Trainable params: 167,324\n",
      "Non-trainable params: 736\n",
      "__________________________________________________________________________________________________\n",
      "\n",
      "[[0.24977717 0.2506411  0.24906778 0.25051397]\n",
      " [0.25026715 0.25101098 0.24840541 0.25031644]\n",
      " [0.2499757  0.25117937 0.24847005 0.25037494]\n",
      " ...\n",
      " [0.25034598 0.25006825 0.24985637 0.24972945]\n",
      " [0.25034064 0.2498511  0.24997747 0.24983081]\n",
      " [0.250155   0.25004098 0.24983251 0.24997151]]\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "\n",
    "\"\"\"\n",
    "To simplify the model i m creating some custom layers\n",
    "\n",
    " - the ConvPassLayer to make a convolution with the appropriate padding and activation as described in the paper\n",
    " - DownPassLayer to make a max pooling and 2 convolutions\n",
    " - UpPassLayer which makes a 'deconvolution', concatenates with same-height layer and makes 2 conv\n",
    "\n",
    "\"\"\"\n",
    "class ConvPassLayer(layers.Layer):\n",
    "    def __init__(self, n_conv_filters, activation='relu', padding_size = 4, **kwargs):\n",
    "        super(ConvPassLayer, self).__init__(**kwargs)\n",
    "        # parameters\n",
    "        self.n_conv_filters = n_conv_filters\n",
    "        self.activation = activation\n",
    "        self.padding_size = padding_size\n",
    "        # layers\n",
    "        self.zero_padding = ZeroPadding1D(padding=self.padding_size)\n",
    "        self.conv = Conv1D(filters = self.n_conv_filters, kernel_size = 9)\n",
    "        self.batch_norm = BatchNormalization()\n",
    "        self.activation_layer = Activation(self.activation)\n",
    "\n",
    "    def build(self, input_shape):\n",
    "        self.input_layer = input_shape\n",
    "\n",
    "    def get_config(self):\n",
    "        config = super().get_config()\n",
    "        config.update({\n",
    "            \"n_conv_filters\": self.n_conv_filters,\n",
    "            \"activation\": self.activation,\n",
    "            \"padding_size\": self.padding_size\n",
    "        })\n",
    "        return config\n",
    "\n",
    "    def call(self, x):\n",
    "        # apply 4 padding in 1 dimension\n",
    "        pad1 = self.zero_padding(x)\n",
    "        # apply a convolution with kernel size 9\n",
    "        conv1 = self.conv(pad1)\n",
    "        # apply batch normalization\n",
    "        bn1 = self.batch_norm(conv1)\n",
    "        # apply activation function\n",
    "        act1 = self.activation_layer(bn1)\n",
    "        return act1\n",
    "\n",
    "    def compute_output_shape(self, input_shape):\n",
    "        return input_shape\n",
    "\n",
    "class DownPassLayer(layers.Layer):\n",
    "    def __init__(self, n_conv_filters, activation='relu', padding_size = 4, **kwargs):\n",
    "        super(DownPassLayer, self).__init__(**kwargs)\n",
    "        # parameters\n",
    "        self.n_conv_filters = n_conv_filters\n",
    "        self.activation = activation\n",
    "        self.padding_size = padding_size\n",
    "        # layers\n",
    "        self.max_pool = MaxPooling1D(pool_size = 2)\n",
    "        self.conv_pass_layer1 = ConvPassLayer(self.n_conv_filters, self.activation, self.padding_size)\n",
    "        self.conv_pass_layer2 = ConvPassLayer(self.n_conv_filters, self.activation, self.padding_size)\n",
    "\n",
    "    def build(self, input_shape):\n",
    "        self.input_layer = input_shape\n",
    "\n",
    "    def call(self, x):\n",
    "        x = self.max_pool(x)\n",
    "        x = self.conv_pass_layer1(x)\n",
    "        x = self.conv_pass_layer2(x)\n",
    "        return x\n",
    "\n",
    "    def compute_output_shape(self, input_shape):\n",
    "        return input_shape//2\n",
    "    \n",
    "    def get_config(self):\n",
    "        config = super().get_config()\n",
    "        config.update({\n",
    "            \"n_conv_filters\": self.n_conv_filters,\n",
    "            \"activation\": self.activation,\n",
    "            \"padding_size\": self.padding_size\n",
    "        })\n",
    "        return config\n",
    "\n",
    "class UpPassLayer(layers.Layer):\n",
    "    def __init__(self, n_conv_filters, activation='relu', padding_size = 4, **kwargs):\n",
    "        super(UpPassLayer, self).__init__(**kwargs)\n",
    "        # parameters\n",
    "        self.n_conv_filters = n_conv_filters\n",
    "        self.activation = activation\n",
    "        self.padding_size = padding_size\n",
    "        # layers\n",
    "        # applys two convolution and pooling layers\n",
    "        self.deconv = Conv1DTranspose(filters = self.n_conv_filters, kernel_size = 8, strides = 2, padding = 'same')\n",
    "        self.concat = Concatenate()\n",
    "        self.conv_pass_layer1 = ConvPassLayer(self.n_conv_filters//2, self.activation, self.padding_size)\n",
    "        self.conv_pass_layer2 = ConvPassLayer(self.n_conv_filters//2, self.activation, self.padding_size)\n",
    "\n",
    "    def build(self, input_shape):\n",
    "        self.input_layer = input_shape\n",
    "\n",
    "    def call(self, same_layer, lower_layer):\n",
    "        deconv = self.deconv(lower_layer)\n",
    "        conc = self.concat([same_layer, deconv])\n",
    "        conv1 = self.conv_pass_layer1(conc)\n",
    "        conv2 = self.conv_pass_layer2(conv1)\n",
    "        \n",
    "        return conv2\n",
    "\n",
    "\n",
    "\n",
    "    def get_config(self):\n",
    "        config = super().get_config()\n",
    "        config.update({\n",
    "            \"n_conv_filters\": self.n_conv_filters,\n",
    "            \"activation\": self.activation,\n",
    "            \"padding_size\": self.padding_size\n",
    "        })\n",
    "        return config\n",
    "\n",
    "\n",
    "def UNET(inpusize =ECG_SIZE):\n",
    "    # first block\n",
    "    inputs = Input((inpusize, 1))\n",
    "    block1layer1 = ConvPassLayer(4)(inputs)\n",
    "    block1layer2 = ConvPassLayer(4)(block1layer1)\n",
    "\n",
    "    # convolution blocks\n",
    "    block2layer2 = DownPassLayer(8)(block1layer2)\n",
    "    block3layer2 = DownPassLayer(16)(block2layer2)\n",
    "    block4layer2 = DownPassLayer(32)(block3layer2)\n",
    "\n",
    "    # 'bottleneck' layer\n",
    "    pool4 = MaxPooling1D(2)(block4layer2)\n",
    "    block5layer1 = ConvPassLayer(64)(pool4)\n",
    "    block5layer2 = ConvPassLayer(64)(block5layer1)\n",
    "\n",
    "    # deconv and concatenation\n",
    "    block4up = UpPassLayer(64)(block4layer2, block5layer2)\n",
    "    block3up = UpPassLayer(32)(block3layer2, block4up)\n",
    "    block2up = UpPassLayer(16)(block2layer2, block3up)\n",
    "\n",
    "    # final block\n",
    "    transposed1 = Conv1DTranspose(filters = 8, kernel_size = 8, strides = 2, padding = 'same')(block2up)\n",
    "    block1layer3 = Concatenate()([transposed1, block1layer2])\n",
    "    block1layer4 = ConvPassLayer(4)(block1layer3)\n",
    "    block1layer5 = ConvPassLayer(4, activation='softmax')(block1layer4)\n",
    "\n",
    "    # create model\n",
    "    return Model(inputs, block1layer5)\n",
    "\n",
    "# just plot model and a prediction\n",
    "# to check that layer sizes matches\n",
    "m = UNET()\n",
    "m.summary()\n",
    "print(f\"\"\"\n",
    "{m.predict(X[0].reshape(1, ECG_SIZE, 1))[0]}\n",
    "\"\"\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now the program load the model if already saved (also test-set is loaded)\n",
    "\n",
    "If the model is not saved (present in the current direcory) than a new one is going to be trained and saved"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded model from trained_model_656.h5\n",
      "92/92 [==============================] - 9s 72ms/step - loss: 0.1112\n",
      "Test accuracy: 0.11124581843614578\n"
     ]
    }
   ],
   "source": [
    "# check if the model is already in the direcotry\n",
    "saved_model_name = f\"trained_model_{ECG_SIZE}.h5\"\n",
    "if file_exists(saved_model_name):\n",
    "    model = load_model(saved_model_name,\n",
    "        custom_objects={\n",
    "            'ConvPassLayer': ConvPassLayer,\n",
    "            'DownPassLayer': DownPassLayer,\n",
    "            'UpPassLayer': UpPassLayer\n",
    "            }\n",
    "    )\n",
    "    X_test = np.load(f\"X_test_{ECG_SIZE}.npy\")\n",
    "    Y_test = np.load(f\"Y_test_{ECG_SIZE}.npy\")\n",
    "    print(f\"Loaded model from {saved_model_name}\")\n",
    "    print(f\"Test accuracy: {model.evaluate(X_test, Y_test)}\")\n",
    "else:\n",
    "\n",
    "    m = UNET()\n",
    "\n",
    "    # create an optimizer\n",
    "    opt = Adam(learning_rate=0.0005)\n",
    "    m.compile(optimizer=opt, loss='binary_crossentropy') \n",
    "\n",
    "    # split the data in train and test\n",
    "    X_train, X_test, Y_train, Y_test = train_test_split(\n",
    "        np.array(X),#X,\n",
    "        np.array(Y, dtype=np.float32),#Y,\n",
    "        test_size=0.31)\n",
    "\n",
    "    train_history = m.fit(\n",
    "        X_train, Y_train,\n",
    "        epochs=15,\n",
    "        batch_size=8,\n",
    "        validation_data=(X_test, Y_test)\n",
    "    )\n",
    "\n",
    "    m.save( # https://www.tensorflow.org/guide/keras/save_and_serialize\n",
    "        f\"trained_model_{ECG_SIZE}.h5\"\n",
    "    )\n",
    "    np.save(f\"X_test_{ECG_SIZE}.npy\", X_test)\n",
    "    np.save(f\"Y_test_{ECG_SIZE}.npy\", Y_test)\n",
    "\n",
    "    plt.figure()\n",
    "    plt.plot(train_history.history[\"loss\"], label=\"train loss\")\n",
    "    plt.plot(train_history.history[\"val_loss\"], label=\"val loss\")\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Measurements\n",
    "\n",
    "Model scorec can be quite tricky to measure and the papar is not very detailed on how to handle certain situation, i took a simple approach... probably not the same, but quite close."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "```\n",
    "Frequenza ecg nel dataset = 500hz\n",
    "-> 500campioni/sec\n",
    "-> 5camp/10ms (5campioni durano un tempo di 10ms)\n",
    "-> 1passo = 2ms \n",
    "-> 75 campioni coprono un arco di 150ms\n",
    "\n",
    "Se la distanza tra l'annotazione del modello e quella del dottore non distano piu di 75 valori\n",
    "allora posso dire che il modello ha annotato bene\n",
    "\n",
    "TP = tutti quelli nella tolleranza (75campioni)\n",
    "FP = se l'algoritmo annota, ma non c'è nulla (di annotato) nella tolleranza\n",
    "FN = se l algoritmo non annota ma c'è qualcosa\n",
    "\n",
    "Per ogni tipologia di onda\n",
    "    Per ogni onda nei dati\n",
    "        per ogni onset/offset\n",
    "            per ogni onda\n",
    "                media e std dev\n",
    "                recall = TP/(TP + FN)\n",
    "                precision= TP/(TP + FP)\n",
    "                f1 = 2 * (recall*precision) / (recall+precision) \n",
    "\n",
    "```\n",
    "\n",
    "C'è però un problema, per quanto il modello sia adestrato si dati che presentatno le wave come contigue, io non ho garanzie che lo siano effettivamente anche nella previsione (perche ne ione il paper facciamo post processing, per esemio per cancellare i segmenti palesemente troppo corti). Questo significa che la previsione del modello potrebbe essere piena di mini-wave chiaramente errori che però io devo contare come Falsi Positivi o Falsi Negativi. Esempio: c'è un ecg che ha il complesso QRS tra il tempo 100 (onset) e il 150 (offset); la previsione identifica però due onde QRS, una tra 90 e 110, l'altra tra 140 e 150, diventa tosto decidere quale onda considerare quella che realmente rappresenta la realtà"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y_test_pred = model.predict(X_test)\n",
    "toll = 75"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "checking P\n",
      "checking QRS\n",
      "checking T\n"
     ]
    }
   ],
   "source": [
    "def find_wave_onset(wave_category:list) -> np.ndarray:\n",
    "    onsets = []\n",
    "    prev = None\n",
    "    for i, val in enumerate(wave_category):\n",
    "        if val != 0 and prev == 0:\n",
    "            onsets.append(i)\n",
    "        prev = val\n",
    "    return np.array(onsets)\n",
    "\n",
    "def find_wave_offset(wave_category:list) -> np.ndarray:\n",
    "    offsets = []\n",
    "    prev = None\n",
    "    for i, val in enumerate(wave_category):\n",
    "        if val == 0 and prev != 0:\n",
    "            offsets.append(i)\n",
    "        prev = val\n",
    "    return np.array(offsets)\n",
    "\n",
    "def get_probably_matching_timepoints(shorter:np.ndarray, longer:np.ndarray):\n",
    "    indices_minimizind_distancies = [\n",
    "        np.argmin(row) for row in np.abs(np.subtract.outer(shorter, longer))\n",
    "    ]\n",
    "    return longer[indices_minimizind_distancies]\n",
    "\n",
    "\n",
    "def get_true_false_errs_from_mismatching_timepoints(shorter:np.ndarray, longer:np.ndarray, toll:int):\n",
    "    if len(shorter) == 0:\n",
    "        return len(longer), 0, []\n",
    "\n",
    "    falses = len(longer) - len(shorter)\n",
    "\n",
    "    matched_longer = get_probably_matching_timepoints(shorter, longer)\n",
    "\n",
    "    dists = np.abs(matched_longer - shorter)\n",
    "\n",
    "    trues = np.sum(dists <= toll)\n",
    "    errs = list(dists[dists < toll])\n",
    "\n",
    "    return falses, trues, errs \n",
    "\n",
    "\n",
    "WAVE_TO_COLUMN = {\n",
    "    \"None\": 0,\n",
    "    \"P\":3,\n",
    "    \"QRS\":1,\n",
    "    \"T\":2,\n",
    "}\n",
    "\n",
    "WAVE_DATA = {\n",
    "    \"None\": None,\n",
    "    \"P\": {},\n",
    "    \"QRS\": {},\n",
    "    \"T\": {},\n",
    "}\n",
    "\n",
    "\n",
    "\n",
    "for wave_name, column_index in WAVE_TO_COLUMN.items():\n",
    "    if column_index == 0:\n",
    "        continue\n",
    "    print(\"checking\", wave_name)\n",
    "    for phase, time_point_function in zip([\"onset\", \"offset\"],[find_wave_onset, find_wave_offset]):\n",
    "\n",
    "        evaluation = {\n",
    "            \"TP\": 0,\n",
    "            \"FP\": 0,\n",
    "            \"FN\": 0,\n",
    "            \"errors\": [],\n",
    "            \"mean_iou\": MeanIoU(num_classes=4)\n",
    "        }\n",
    "\n",
    "        for y_test, y_pred in zip(Y_test, Y_test_pred):\n",
    "            wave_test = (np.argmax(y_test, axis=1) == column_index).astype(int)\n",
    "            wave_pred = (np.argmax(y_pred, axis=1) == column_index).astype(int)\n",
    "\n",
    "            wave_test_points = time_point_function(wave_test)\n",
    "            wave_pred_points = time_point_function(wave_pred)\n",
    "            try:\n",
    "                if len(wave_test_points) == len(wave_pred_points):\n",
    "                    # i ASSUME that wave match and that there might be true positives\n",
    "                    # still have to if correspoding points are in the tollerance\n",
    "                    _, trues, errs = get_true_false_errs_from_mismatching_timepoints(wave_test_points, wave_pred_points, toll)\n",
    "                    evaluation[\"TP\"] += trues\n",
    "                    evaluation[\"errors\"] += errs\n",
    "                    \n",
    "                elif len(wave_test_points) > len(wave_pred_points):\n",
    "                    # if 'wave_test_points' is longer i have false negatives\n",
    "                    falses, trues, errs = get_true_false_errs_from_mismatching_timepoints(wave_pred_points, wave_test_points, toll)\n",
    "                    evaluation[\"TP\"] += trues\n",
    "                    evaluation[\"FN\"] += falses\n",
    "                    evaluation[\"errors\"] += errs\n",
    "\n",
    "                elif len(wave_test_points) < len(wave_pred_points):\n",
    "                    # if 'wave_pred_points' is longer i have false positives\n",
    "                    falses, trues, errs = get_true_false_errs_from_mismatching_timepoints(wave_test_points, wave_pred_points, toll)\n",
    "\n",
    "                    evaluation[\"TP\"] += trues\n",
    "                    evaluation[\"FP\"] += falses\n",
    "                    evaluation[\"errors\"] += errs\n",
    "            except Exception as e:\n",
    "                print(wave_test_points, wave_pred_points, len(wave_test_points), len(wave_pred_points))\n",
    "                raise e\n",
    "\n",
    "\n",
    "            evaluation[\"mean_iou\"].update_state(wave_test, wave_pred)\n",
    "\n",
    "        \n",
    "        WAVE_DATA[wave_name][phase] = evaluation\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "P\n",
      "    onset\n",
      "        recall = 95.37\n",
      "        precision = 92.19\n",
      "        13.62 +- 15.81 (ms)\n",
      "        m_iou = 85.51\n",
      "    offset\n",
      "        recall = 98.58\n",
      "        precision = 94.75\n",
      "        6.38 +- 10.81 (ms)\n",
      "        m_iou = 85.51\n",
      "QRS\n",
      "    onset\n",
      "        recall = 97.75\n",
      "        precision = 98.12\n",
      "        6.23 +- 6.66 (ms)\n",
      "        m_iou = 91.83\n",
      "    offset\n",
      "        recall = 99.21\n",
      "        precision = 98.33\n",
      "        5.39 +- 8.15 (ms)\n",
      "        m_iou = 91.83\n",
      "T\n",
      "    onset\n",
      "        recall = 97.45\n",
      "        precision = 93.01\n",
      "        18.85 +- 18.83 (ms)\n",
      "        m_iou = 87.6\n",
      "    offset\n",
      "        recall = 98.84\n",
      "        precision = 94.72\n",
      "        10.18 +- 17.58 (ms)\n",
      "        m_iou = 87.6\n"
     ]
    }
   ],
   "source": [
    "for wave, phase_data in WAVE_DATA.items():\n",
    "    if wave == \"None\":\n",
    "        continue\n",
    "    else:\n",
    "        print(f\"{wave}\")\n",
    "\n",
    "    for phase, data in phase_data.items():\n",
    "        err = np.array(data['errors']) * 2\n",
    "        rec = data['TP']/(data['TP'] + data['FN'])\n",
    "        prec = data['TP']/(data['TP'] + data['FP']) # TP/(TP + FP)\n",
    "        miou = data[\"mean_iou\"].result().numpy()\n",
    "        print(f\"\"\"    {phase}\n",
    "        recall = {round(rec * 100,2)}\n",
    "        precision = {round(prec * 100,2)}\n",
    "        {err.mean().round(2)} +- {err.std().round(2)} (ms)\n",
    "        m_iou = {round(miou * 100, 2)}\"\"\" )\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Nice plots\n",
    "\n",
    "Some art "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "ecg_index = 11"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Fist the ECG signal with the right annotations "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot categories on ecg\n",
    "signal = X_test[ecg_index]\n",
    "categories = np.argmax(Y_test[ecg_index], axis=1)\n",
    "predicted_categories = np.argmax(Y_test_pred[ecg_index], axis=1)\n",
    "\n",
    "# get time point where category changes\n",
    "time_points = np.where(categories[:-1] != categories[1:])[0]\n",
    "\n",
    "# chek if category was at beginning or end\n",
    "if categories[0] != 0:\n",
    "    # if category was at beginning, add first time point\n",
    "    time_points = np.append(0, time_points)\n",
    "if categories[-1] != 0:\n",
    "    # if category was at end, add last time point\n",
    "    time_points = np.append(time_points, len(categories))\n",
    "\n",
    "# plot the signal\n",
    "plt.figure(figsize=(10, 4))\n",
    "plt.plot(signal)\n",
    "for b,e in grouped(time_points, 2):\n",
    "    plt.axvspan(b, e, facecolor='0.5', alpha=0.5)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "And then the same ECG with predicted segmentation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "signal = X_test[ecg_index]\n",
    "categories = np.argmax(Y_test[ecg_index], axis=1)\n",
    "predicted_categories = np.argmax(Y_test_pred[ecg_index], axis=1)\n",
    "\n",
    "# get time point where category changes\n",
    "time_points = np.where(predicted_categories[:-1] != predicted_categories[1:])[0]\n",
    "\n",
    "\n",
    "# chek if category was at beginning or end\n",
    "if predicted_categories[0] != 0:\n",
    "    # if category was at beginning, add first time point\n",
    "    time_points = np.append(0, time_points)\n",
    "if predicted_categories[-1] != 0:\n",
    "    # if category was at end, add last time point\n",
    "    time_points = np.append(time_points, len(predicted_categories))\n",
    "\n",
    "# plot the signal\n",
    "plt.figure(figsize=(10, 4))\n",
    "plt.plot(signal)\n",
    "for b,e in grouped(time_points, 2):\n",
    "    plt.axvspan(b, e, facecolor='0.5', alpha=0.5)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = Y_test_pred[ecg_index]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x2086cda8490>"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 4))\n",
    "\n",
    "plt.plot(signal, color = \"gray\")\n",
    "plt.plot(pred[:,0]*signal, alpha=0.5) # predicted not interesting\n",
    "\n",
    "plt.plot(pred[:,3]*signal, alpha=0.9, label= \"p\", c=SEGMENT_TO_COLOR['p']) # predicted p\n",
    "\n",
    "plt.plot(pred[:,1]*signal, alpha=0.9, label= \"qrs\", c=SEGMENT_TO_COLOR['N']) # predicted qrs\n",
    "plt.plot(pred[:,2]*signal, alpha=0.9, label= \"t\", c=SEGMENT_TO_COLOR['t']) # predicted t\n",
    "\n",
    "plt.legend(\n",
    "    loc='upper right',\n",
    "    ncol=1,\n",
    "\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Lastly, alwasys the same ECG but colored depending on the probability that the model assign to every sample of being of a certainc category"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(-0.6420765027322405, 0.24153005464480876)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from matplotlib.collections import LineCollection\n",
    "from matplotlib.colors import ListedColormap\n",
    "\n",
    "x   = np.arange(len(signal)) \n",
    "t = np.linspace(0,1,x.shape[0]) # your \"time\" variable\n",
    "\n",
    "lightblue  = [(0.3, 0.5, 0.9, 0.99 * p) for p in pred[:,1]]\n",
    "lightred   = [(0.99, 0.6, 0.5, 0.99 * p) for p in pred[:,2]]\n",
    "lightgreen = [(0.55, 0.99, 0.66, 0.99 * p) for p in pred[:,3]]\n",
    "\n",
    "\n",
    "# set up a list of (x,y) points\n",
    "points = np.array([x,signal]).transpose().reshape(-1,1,2)\n",
    "segs = np.concatenate([points[:-1],points[1:]],axis=1)\n",
    "\n",
    "# make the collection of segments\n",
    "lcblue = LineCollection(segs, cmap=ListedColormap( lightblue))\n",
    "lcblue.set_array(t)\n",
    "lcred = LineCollection(segs, cmap=ListedColormap( lightred))\n",
    "lcred.set_array(t)\n",
    "lcgreen = LineCollection(segs, cmap=ListedColormap( lightgreen))\n",
    "lcgreen.set_array(t)\n",
    "\n",
    "\n",
    "# plot the collection\n",
    "plt.figure(figsize=(15, 5))\n",
    "plt.plot(signal, alpha = 0.4, linewidth = 2, color='gray')\n",
    "plt.gca().add_collection(lcblue) # add the collection to the plot\n",
    "plt.gca().add_collection(lcred) # add the collection to the plot\n",
    "plt.gca().add_collection(lcgreen) # add the collection to the plot\n",
    "plt.xlim(x.min(), x.max()) # line collections don't auto-scale the plot\n",
    "plt.ylim(signal.min(), signal.max())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Differencies\n",
    "\n",
    "With respect to the original paper\n",
    "\n",
    "- not sure we used the same dataset\n",
    "    - the authors used a 455 ECG dataset (they name it \"extended LUDB\", but i couldn't find the extended version)\n",
    "- i do not do preprocessing\n",
    "    - articles's dataset is made indipendent from sample frequency and signal length\n",
    "- we train the model on different length segment\n",
    "    - paper takes (from every signal) two 4sec long cuts, I took 4 segments each one 1sec long \n",
    "- i did much less experiments to evaluate the model\n",
    "    - for example: the paper also show results for segmentation quality made just with \"II\" lead\n",
    "- i measured TP, FN, FP in a different way\n",
    "    - the paper is not very detailed on how to handle \"complicated\" and \"inaccurate\" prediction, the model can produce both FN and FP in the same prediction and it is hart to account fot that during measurements; the decided to simplify my job making some assumption (eg. if there are the same number of onset/offset the model has no FN/FP, if there are more wave in the true ECG then there are only FN), still results seems to look like the ones on the paper (P wave are ok, QRS most easilly detectd, T wave with largest uncertenty)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "896b573f4b38a987c3892cea291473bfed7314c3532c9a0783074b5d4a7e8998"
  },
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.8.2"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}