🏥 Patients Management System

🎯 A full-stack Patient Management + Medical Insurance Cost Prediction system built with FastAPI and Machine Learning — manages patient records via REST API and predicts insurance costs using a trained ML model.

---

🎯 Problem Statement

Healthcare systems need two things simultaneously:

• ❌ A reliable way to manage patient records (add, update, delete, view)
• ❌ A tool to predict medical insurance costs based on patient profile

This system solves both — combining a FastAPI-powered CRUD backend with a trained ML regression model served as a live prediction endpoint.

---

✨ Key Features

Feature	Description
👥 Patient CRUD	Add, view, update, delete patient records via REST API
🔍 Patient Lookup	Fetch individual patient by ID
💰 Insurance Prediction	Predict medical cost from patient profile using ML
🧠 Trained ML Model	Pre-trained model served via FastAPI endpoint
✅ Data Validation	Pydantic models for strict input validation
📄 Auto API Docs	Swagger UI at /docs out of the box

---

🔄 System Architecture

Patient Data (JSON)
      ↓
FastAPI REST API  ←→  Pydantic Validation
      ↓
  Two Services:
  ┌─────────────────────────────┐
  │  1. Patient Management      │
  │     CRUD Operations         │
  │     patients.json storage   │
  └─────────────────────────────┘
  ┌─────────────────────────────┐
  │  2. Insurance Predictor     │
  │     model.pkl (ML Model)    │
  │     insurance.csv dataset   │
  └─────────────────────────────┘

---

💰 ML Model — Insurance Cost Prediction

Trained on the Medical Insurance Cost Dataset to predict a patient's insurance charges based on personal health profile.

Input Features:

Feature	Type	Description
age	int	Patient age in years
sex	str	Gender (male/female)
bmi	float	Body Mass Index
children	int	Number of dependents
smoker	str	Smoking status (yes/no)
region	str	Geographic region

Output: Predicted annual insurance cost (USD)

---

🔌 API Endpoints

GET    /                          # Welcome / health check
GET    /patients                  # Get all patients
GET    /patients/{patient_id}     # Get patient by ID
POST   /patients                  # Add new patient
PUT    /patients/{patient_id}     # Update patient record
DELETE /patients/{patient_id}     # Delete patient

POST   /predict                   # Predict insurance cost (ML)

Example — Insurance Prediction

POST /predict
Content-Type: application/json

{
  "age": 35,
  "sex": "male",
  "bmi": 27.5,
  "children": 2,
  "smoker": "no",
  "region": "southwest"
}

# Response:
{
  "predicted_insurance_cost": 5842.32
}

---

🛠️ Tech Stack

Layer	Technology
API Framework	FastAPI
Data Validation	Pydantic v2
ML Model	Scikit-learn (Linear Regression)
Model Storage	Pickle (model.pkl)
Data Storage	JSON file (patients.json)
Dataset	Medical Insurance CSV
Notebook	Jupyter Notebook
Language	Python 3.x

---

🚀 Getting Started

# Clone the repo
git clone https://github.com/tashfeen786/PatientsManagementSystem.git
cd PatientsManagementSystem

# Install dependencies
pip install fastapi uvicorn scikit-learn pandas numpy pydantic

# Start FastAPI server
uvicorn main:app --reload

Visit:

• 🌐 API: http://localhost:8000
• 📖 Swagger Docs: http://localhost:8000/docs
• 📘 ReDoc: http://localhost:8000/redoc

---

🏗️ Project Structure

PatientsManagementSystem/
│
├── main.py                    # FastAPI app — all routes & logic
├── app.py                     # App entry point
├── fastapi_ml_model.ipynb     # ML model training notebook
├── model.pkl                  # Trained ML model (serialized)
├── insurance.csv              # Medical insurance dataset
└── patiants.json              # Patient records storage

---

📓 ML Training Notebook

The fastapi_ml_model.ipynb notebook covers:

• 📊 Exploratory Data Analysis on insurance dataset
• 🔧 Feature engineering & encoding (sex, smoker, region)
• 🤖 Model training — Linear Regression
• 📈 Model evaluation — R² score, MAE, RMSE
• 💾 Model serialization using Pickle → model.pkl
• ⚡ FastAPI integration for serving predictions

---

🧠 Key Concepts Demonstrated

• ✅ REST API Design — proper HTTP methods and status codes
• ✅ ML Model Serving — loading pickle model in FastAPI
• ✅ Pydantic Validation — type-safe request/response models
• ✅ CRUD Operations — full Create, Read, Update, Delete
• ✅ End-to-End ML Pipeline — from training to API deployment
• ✅ JSON-based persistence — lightweight data storage

---

🔮 Future Improvements

• Replace JSON storage with PostgreSQL database
• Add JWT authentication for secure patient access
• Integrate more ML models (XGBoost, Random Forest)
• Add patient history and medical records tracking
• Deploy on Railway / Render
• Add Docker containerization

---

👨‍💻 Author

Tashfeen Aziz — AI/ML Engineer & Python Developer

---

⭐ If you found this project helpful, please give it a star!