Heart Failure Predictor

Heart Failure Readmission Predictor

This project is a machine learning-powered web application designed to predict the likelihood of 30-day hospital readmission for heart failure patients, it enables clinicians to identify high-risk patients early and make informed care decisions.

Team - Codiologist

Demo

• Project Link : https://heart-failure-predictor-brown.vercel.app/
• Video Explanation : https://drive.google.com/file/d/1CLqVrzb2xeWla4cS2JcT6XDaslOB2NZs/view?usp=sharing
• Figma File : https://www.figma.com/design/LQXSo9dh27iJVGmhIfcltt/Landing-page?node-id=0-1&t=diKrBpvhCiMr9QoQ-1
• API Documentation : https://documenter.getpostman.com/view/45294944/2sB2qcCg7z
• ML Model Documentation : https://drive.google.com/file/d/1kImpe-VHfxwSqQu7t3olVJf9ODkin7-F/view?usp=sharing

Features

• Predicts 30-day readmission risk using clinical and lab data
• Clean, responsive UI with modern design
• Medical form with validation and dynamic lab test input
• Real-time prediction via Flask API
• Visual patient report with risk percentage & printable summary
• Google or Email-based Sign-Up/Sign-In via Supabase

ML Model

• Model Type: XGBoost Classifier
• Tuning: Hyperparameter optimization with Optuna
• Inputs:
  • Discharge location
  • Gender
  • Length of stay
  • Abnormal lab count
  • Lab test count
  • ICD9-based severity
• Backend: Flask API hosted on Render
• Frontend: Built using React.js (Next.js) and Tailwind CSS
• Integration: CORS-enabled for seamless frontend-backend communication

How It Works

1. User signs up or signs in via email or Google.
2. Fills in medical form with clinical details.
3. Submits form → data sent to /predict API endpoint.
4. Receives a risk score (0–100%) and a diagnosis report.
5. Can print or download the generated report.

Authentication (Supabase)

This project uses Supabase Auth for secure authentication with Email/Password and Google OAuth providers.

Method	Route	Description
POST	/signup	Register a new user with email/password
POST	/signin	Login user with email and password
POST	/auth/callback	Sign in user with Google OAuth

Database Schema: patientDetail

Column	Type	Description
id	UUID	Primary key for the patient record
userId	UUID	Supabase Auth user ID (foreign key reference)
patientId	TEXT	Custom patient identifier (e.g., PAT1654)
name	TEXT	Full name of the patient
age	INTEGER	Age of the patient
gender	TEXT	Gender of the patient (Male, Female, etc.)
admitDate	DATE	Date when the patient was admitted
dischargeDate	DATE	Date when the patient was discharged
dischargeLocation	TEXT	Discharge destination (e.g., HOSPICE-HOME)
labTests	JSONB	List of lab test results (status and ICD-9 code)
probability	FLOAT	Model-predicted probability (e.g., readmission risk)
created_at	TIMESTAMP	Timestamp when the record was created

Backend API Route – Store Patient Data

Method	Route	Description
POST	/api/store	Create and store a new patient record

Tech Stack

Layer	Technologies
Frontend	Next.js, TailwindCSS, Framer Motion
Backend	Flask, Python
Machine Learning	XGBoost, Optuna, NumPy, Joblib
Authentication	Supabase Auth (Email & Google OAuth)

Deployment

-Backend: Flask API hosted on [Render] (https://heart-failure-predictor-z0j1.onrender.com/predict) -Frontend: Next.js, deployable on [Vercel] (https://heart-failure-predictor-brown.vercel.app/)

Installation & Setup Guide

1. clone the repo

  git clone https://github.com/your-username/heart-failure-prediction.git
  cd heart-failure-prediction

2. Backend Setup (Flask API + XGBoost)

cd backend
python -m venv venv
source venv/bin/activate       # For Windows: venv\Scripts\activate
pip install -r requirements.txt
python app.py  #run the flask api

3. Frontend Setup (Next.js)

cd ../frontend
npm install

4. setup environment variable for frontend

NEXT_PUBLIC_SUPABASE_URL=your_supabase_url
NEXT_PUBLIC_SUPABASE_ANON_KEY=your_supabase_anon_key

5. Start the development server:

npm run dev

6. Access the application at http://localhost:3000.

License

This project is licensed under the MIT License.