Heart failure prediction using Machine Learning

Welcome to the Heart Failure Prediction project! This project leverages machine learning to predict heart failure based on various health metrics.

Overview

This repository contains the code and resources for training and deploying a machine learning model to predict heart failure. The project includes both a frontend application built with React and a backend API built with Python (Flask).

Web app is responsive and tested using various devices.

Features

• Machine Learning Model: SVM with RBF kernel, optimized using grid search cross-validation.
• Frontend: React application for user interaction and visualization.
• Backend: Flask API for serving the machine learning predictions.
• Dockerized Deployment: Easily deploy the application using Docker.

Machine Learning framework

Model selection

After conducting the EDA and an exhaustive grid search cross-validation among three models, we selected an SVM model with C=1, gamma=0.1, and an RBF kernel for deployment. This model uses One-Hot Encoding for categorical features and excludes the RestingBP feature (more info at EDA).

Model Evaluation Results

• Accuracy: 0.875
• F1 Score: 0.8889
• Precision: 0.8762
• Recall: 0.9020
• ROC-AUC: 0.9255

Class	Precision	Recall	F1-Score
No HeartDisease	0.87	0.84	0.86
HeartDisease	0.88	0.90	0.89
Accuracy			0.88
Macro Avg	0.87	0.87	0.87
Weighted Avg	0.87	0.88	0.87

Installation Instructions

Prerequisites

Ensure you have Docker and Docker Compose installed on your machine.

Step 1: Clone the Repository

Clone the repository to your local machine:

git clone https://github.com/gbatsis/heart-failure-predictor.git
cd heart-failure-predictor

Step 2: Ensure the directory structure looks like this:

heart-failure-predictor/

├── backend/

│ ├── Dockerfile

│ ├── requirements.txt

│ └── ... (other backend files)

├── frontend/

│ ├── Dockerfile

│ ├── package.json

│ ├── package-lock.json

│ └── ... (other frontend files)

├── docker-compose.yml

└── ... (other project files)

Step 3: Docker Compose Configuration

Verify the docker-compose.yml file in the root of the project:

services:
  frontend:
    build:
      context: ./frontend
      dockerfile: Dockerfile
    ports:
      - "3000:3000"
    environment:
      - CHOKIDAR_USEPOLLING=true
    volumes:
      - ./frontend:/app
      - /app/node_modules

  backend:
    build:
      context: ./backend
      dockerfile: Dockerfile
    ports:
      - "5050:5050"
    volumes:
      - ./backend:/app

volumes:
  backend_db:

Adjust the ports based on your needs.

Step 4: Build and Run the Docker Containers

Run the following command to build and start the Docker containers:

docker-compose up --build

Issues

1. Message: "docker.errors.DockerException: Error while fetching server API version: ('Connection aborted.', PermissionError(13, 'Permission denied'))" Solution:

sudo docker-compose up --build

2. Sometimes, Docker can use a cached layer that didn't install the packages properly. You can force Docker to rebuild the images without using cache:

  docker-compose build --no-cache
  docker-compose up

Run the project without using Docker:

Step 1: Prepare backend

Create virtual enviroment and install project requirements:

  cd backend
  python3.8 -m venv .venv
  pip install requirements.txt

Run application:

  python app.py 

Step 2: Prepare frontend

Ensure that Node.js is installed on your computer

Start node:

  cd frontend
  npm start

Step 5: Access the Application

Once the containers are running, you can access the application:

Frontend: Open your browser and navigate to http://localhost:3000

Backend: The backend server will be running at http://localhost:5050