https://youtu.be/NVT3C7X6870

MyGraine - Personalized migraine prevention, detection and treatment

An intelligent mobile application that combines health tracking, AI-powered insights, and machine learning to help users prevent and manage migraines effectively.

---

🎯 Project Overview

MyGraine is a comprehensive migraine prevention and management system that integrates multiple data sources to provide personalized insights and predictions. The platform consists of:

Unlike traditional ML systems, MyGraine's model learns and improves in real-time from user feedback, making it increasingly personalized and accurate with continued use.

---

✨ Core Features

📱 Mobile Application

1. Comprehensive Health Tracking

• Manual Tracking: Daily meals, hydration, alcohol consumption
• Device Metrics: Steps, sleep quality, heart rate, screen time, screen brightness, outdoor brightness, usage accuracy
• External Sources: Calendar integration (stress from meetings), weather conditions, barometric pressure

2. Smart Dashboard

• Real-time health status overview with color-coded indicators (Critical → Excellent)
• Pattern detection with automatic warnings for concerning trends
• Integration with multiple data sources for holistic health view
• Dark mode support throughout the app

3. Migraine Tracking & Logging

• Quick migraine entry with intensity slider (1-5 scale)
• Symptom selection (aura, nausea, vomiting, light/sound sensitivity, etc.)
• Duration tracking and personal notes
• AI-powered pattern analysis and insights

4. AI-Powered Insights (Gemini Integration)

• Personalized health analysis based on all tracked metrics
• Trigger pattern identification
• Preventive recommendations
• Real-time generation of actionable insights
• Stylish loading screens with dismiss functionality

5. Comprehensive Analytics

• 30-day trend visualization with line charts
• Migraine frequency and severity tracking
• Sleep quality, stress levels, and hydration correlations
• Export reports for healthcare professionals

6. Emergency Support

• Quick access to migraine relief tips
• Immediate help button on home screen
• Emergency contact information

7. Seamless Onboarding

• Three-step guided setup process
• Personal information (name, age bracket)
• Integration selection (health, device, external sources)
• Immediate dashboard population with realistic mock data

🧠 Backend & Machine Learning

Real-Time Model Training

• Online Learning: Model performs backward passes on user feedback instantly
• Experience Replay: 100-sample buffer prevents catastrophic forgetting
• Confidence-Based Updates: Only updates when prediction confidence < 80%
• Auto-Save: Periodic model checkpoints to Google Cloud Storage

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                     Mobile App (Expo)                       │
│  ┌──────────────┐  ┌──────────────┐  ┌─────────────────┐  │
│  │   Dashboard  │  │   Tracking   │  │   Analytics     │  │
│  │   Warnings   │  │   AI Insights│  │   Reports       │  │
│  └──────────────┘  └──────────────┘  └─────────────────┘  │
│           │                │                    │           │
│           └────────────────┴────────────────────┘           │
└───────────────────────────┬─────────────────────────────────┘
                            │ HTTP/REST
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                   Backend API (FastAPI)                     │
│  ┌──────────────┐  ┌──────────────┐  ┌─────────────────┐  │
│  │  Prediction  │  │  Online      │  │   Gemini AI     │  │
│  │  Service     │  │  Learning    │  │   Integration   │  │
│  └──────────────┘  └──────────────┘  └─────────────────┘  │
│           │                │                    │           │
│           └────────────────┴────────────────────┘           │
└───────────────────────────┬─────────────────────────────────┘
                            │
                            ▼
┌─────────────────────────────────────────────────────────────┐
│                   Google Cloud Platform                     │
│  ┌──────────────┐  ┌──────────────┐  ┌─────────────────┐  │
│  │  Vertex AI   │  │  Cloud Run   │  │   Cloud Storage │  │
│  │  (Training)  │  │  (Serving)   │  │   (Models/Data) │  │
│  └──────────────┘  └──────────────┘  └─────────────────┘  │
└─────────────────────────────────────────────────────────────┘

🛠️ Technology Stack

Mobile

• Framework: React Native (Expo SDK 51)
• Language: TypeScript
• Navigation: Expo Router (file-based)
• State Management: React Context API
• UI Components: Custom + Expo Icons
• Charts: Custom line chart implementation
• HTTP Client: Fetch API

Backend

• Framework: FastAPI (Python 3.9+)
• ML Framework: PyTorch
• Data Processing: Pandas, NumPy, scikit-learn
• API Validation: Pydantic
• Async Runtime: uvicorn
• AI Integration: Google Gemini API

Cloud & DevOps

• Cloud Platform: Google Cloud Platform
• Training: Vertex AI
• Serving: Cloud Run (containerized)
• Storage: Google Cloud Storage
• CI/CD: GitHub Actions (optional)
• Monitoring: Cloud Logging & Monitoring

---

🎓 Use Cases

For Patients

• Track health metrics easily
• Identify migraine triggers
• Get personalized AI recommendations
• Share reports with doctors
• Prevent migraines proactively

For Healthcare Providers

• Review comprehensive 30-day reports
• Analyze patient patterns
• Data-driven treatment decisions
• Track medication effectiveness
• Remote patient monitoring

For Researchers

• Collect anonymized migraine data
• Study trigger patterns at scale
• Validate ML models
• Improve classification accuracy
• Contribute to migraine research

Built with ❤️ for the Aava & Pfizer Hackathon Challenge