Breast Cancer Recurrence Prediction API

This FastAPI-based machine learning API predicts breast cancer recurrence based on structured clinical features. It uses a versioned model architecture to support future model upgrades without breaking compatibility.

🚀 Features

• 🔐 JWT-secured endpoints
• 📊 XGBoost-based recurrence classification
• 🔄 Model versioning via /predecir/predict and /predecir/meta
• 🧠 Input validation with categorical encoders
• 📤 Metadata endpoint to expose model expectations
• 📦 Auto-token refresh Postman collection

📁 Project Structure

predecir-api/
├── main.py
├── model/
│   └── v1/
│       ├── model.pkl
│       └── encoders/
├── routers/
│   └── v1/
│       └── prediction.py
├── entity/
│   └── recurrence_input.py
├── auth/
│   └── auth_handler.py
├── utils/
│   └── logging_config.py
├── core/
│   └── config.py
├── postman/
│   └── BreastCancerRecurrenceAPI_v1.postman_collection.json
└── .env.example

🧪 Running Locally

pip install -r requirements.txt
uvicorn main:app --reload

Set up a .env file using .env.example.

🔑 Authentication

• Use /token to get JWT token (username/password in env)
• Pass token in Authorization: Bearer <token> header

📬 Endpoints

• POST /predecir/predict - Make a prediction
• GET /predecir/meta - Get valid field values
• POST /predecir/token - Get access token

📦 Postman

Import the Postman collection from /postman folder. Includes auto-refreshing token setup.

---

📄 License

MIT License (c) 2025 Carealytica