AI Disease Prediction System

🏥 Advanced AI-powered disease prediction based on symptoms

An intelligent medical assistance system that uses machine learning to predict diseases based on user-selected symptoms, powered by Random Forest classification and Google Gemini AI for detailed disease descriptions.

Docs are live at https://mannu.me/disease-detector

🚀 Quick Start

For Users

Visit the web application and start predicting diseases based on your symptoms:

• Select symptoms from 130+ available options
• Get AI-powered disease predictions
• Receive detailed disease information

For Developers

# Clone repository
git clone <repository-url>
cd disease-detector

# Start backend (Terminal 1)
cd backend && ./run.sh dev

# Start frontend (Terminal 2)
cd frontend && uv run python main.py

📊 System Overview

• 🧠 ML Model: Random Forest Classifier with ~95% accuracy
• 🔬 Symptoms: 132 different medical symptoms
• 🏥 Diseases: 41 different medical conditions
• 🤖 AI Integration: Google Gemini for disease descriptions
• 🌐 Interface: Modern Streamlit web application
• ⚡ API: Flask-based REST API

📁 Project Structure

disease-detector/
├── docs/                    # 📚 Comprehensive documentation
├── ml/                      # 🧠 Machine learning components
├── backend/                 # 🔧 Flask API server
├── frontend/                # 🎨 Streamlit web interface
└── README.md               # 📖 This file

📚 Documentation

Comprehensive documentation is available in the docs/ directory:

Document	Description
📋 Documentation Index	Complete documentation overview
🧠 ML Documentation	Machine learning model details
🔧 Backend Documentation	Flask API reference
🎨 Frontend Documentation	Streamlit interface guide
📡 API Documentation	Complete API reference
🚀 Deployment Guide	Production deployment
🛠️ Development Guide	Developer setup and workflow
👤 User Guide	End-user instructions

🎯 Features

Core Features

• Symptom-based Prediction: Select from 130+ medical symptoms
• AI Disease Descriptions: Detailed information powered by Google Gemini
• Modern UI: Responsive, dark-themed web interface
• REST API: Programmatic access to predictions
• Production Ready: Docker support and production configurations

Technical Features

• High Accuracy: ~95% prediction accuracy on test data
• Real-time Processing: Fast symptom analysis and prediction
• Async Operations: Non-blocking disease description fetching
• Error Handling: Comprehensive error management
• Cross-platform: Works on Linux, macOS, and Windows

🔧 Technology Stack

• Machine Learning: scikit-learn, pandas, numpy
• Backend: Flask, Gunicorn, Flask-CORS
• Frontend: Streamlit, aiohttp
• AI Integration: Google Gemini API
• Data Processing: joblib, JSON-based mappings
• Deployment: Docker, production WSGI server

⚡ Quick Examples

API Usage

# Get disease prediction
curl -X POST http://localhost:8000/predict \
  -H "Content-Type: application/json" \
  -d '{"symptoms": ["Itching", "Skin Rash", "High Fever"]}'

# Get disease description
curl -X POST http://localhost:8000/disease_description \
  -H "Content-Type: application/json" \
  -d '{"disease_name": "Common Cold"}'

Python Integration

import requests

# Predict disease
response = requests.post('http://localhost:8000/predict',
    json={"symptoms": ["Headache", "Fever", "Nausea"]})
disease = response.json()['disease']
print(f"Predicted disease: {disease}")

🏥 Supported Conditions

The system can predict 41 different medical conditions including:

Common Conditions: Cold, Flu, Pneumonia, Diabetes, Hypertension
Infectious Diseases: Malaria, Dengue, Typhoid, Hepatitis variants
Chronic Conditions: Arthritis, GERD, Peptic Ulcer Disease
Other Conditions: Migraine, Jaundice, Heart Attack, and more

🛡️ Medical Disclaimer

⚠️ Important: This system is for educational and informational purposes only. It should not be used as a substitute for professional medical advice, diagnosis, or treatment. Always consult with qualified healthcare providers for medical concerns.

🤝 Contributing

We welcome contributions! Please see our Development Guide for:

• Setting up the development environment
• Understanding the codebase
• Contribution guidelines
• Testing procedures

📞 Support

• Documentation: Check the docs/ directory
• Issues: Create a GitHub issue for bugs or feature requests
• Questions: Refer to the User Guide

📈 Performance

• Model Accuracy: ~95% on test dataset
• API Response Time: <500ms for predictions
• Symptoms Supported: 132 different symptoms
• Disease Categories: 41 medical conditions
• Concurrent Users: Supports multiple simultaneous users

🔐 Privacy

• No personal data collection
• Temporary session-based processing
• No medical record storage
• Anonymous usage tracking

---

Built with ❤️ for better healthcare accessibility through AI