Federated Learning for Diabetes Detection

A privacy-preserving system that enables hospitals to collaboratively train diabetes prediction models without sharing patient data.

Features

• 🛡️ Privacy-first: Data never leaves local hospitals
• 🤝 Collaborative: Multiple institutions improve model together
• 🏥 Clinical UI: Doctor-friendly dashboard with explanations
• ⚡ Fast predictions: <500ms inference via ONNX runtime

Quick Start

Prerequisites

• Python 3.9+
• Node.js 18+

Installation

# Clone repo
git clone https://github.com/ManNjoro/federated_learning.git
cd federated_learning

# Set up Python
python -m venv venv
source venv/bin/activate  # Linux/MacOS
# venv\Scripts\activate  # Windows
pip install -r requirements.txt

# Set up frontend
cd frontend
npm install

# Start backend (Flask + Flower)
cd backend
python flower.py

# In new terminal - start frontend
cd frontend
npm run dev

Access: http://localhost:5173 # Your port could be different

🌐 Usage

For Hospitals

• Upload anonymized patient data through the admin interface
• Participate in federated training rounds
• Monitor local model performance

For Clinicians

• Enter patient symptoms through the dashboard
• View real-time diabetes risk predictions
• See explainable AI insights

📸 Screenshots

Dashboard

Patient Data Upload

Model Training

📜 License

MIT License - See LICENSE for details.

📬 Contact

For questions or collaborations, please open an issue or contact [elijohnmwoho@gmail.com].