A full-stack AI application that predicts early-stage diseases using machine learning. Built as my Final Year Project (BS Computer Science).
🔗 Live Demo: predictcare-app.vercel.app
📡 Backend API: Deployed on Render
Users enter their symptoms and health data into the mobile app. The app sends this to a machine learning model which predicts whether the user is at risk for specific diseases — returning results in real time.
| Layer | Technology |
|---|---|
| ML Model | Python · scikit-learn · Gradient Boosting |
| Data Analysis | pandas · NumPy · Google Colab |
| Backend API | FastAPI (Python) |
| Frontend | React Native · TypeScript · Expo |
| Database | MySQL |
| Deployment | Render (backend) · Vercel (frontend) |
- Dataset: 10 years of historical health/disease data (synthetic, representing real-world patterns)
- Best performing model: Gradient Boosting Classifier
- Training accuracy: 98% on training set
- Note: Model was trained on synthetic data due to real-world data availability constraints. Production use would require validation on real patient data and clinical review.
Backend:
cd "backend predictcare"
pip install -r requirements.txt
uvicorn main:app --reloadFrontend:
npm install
npx expo start- Building and evaluating multiple ML models (Random Forest, Gradient Boosting)
- Deploying Python ML models as production REST APIs with FastAPI
- Connecting a mobile frontend to a live ML backend
- Working with synthetic datasets and understanding their limitations