🧠 EEG-Based Depression Detection

A full-stack mobile health application that detects depression from EEG signals using machine learning — built for accessible, early mental health screening. Achieves 90%+ classification accuracy on the MODMA dataset.

---

📌 Overview

Depression affects over 280 million people globally, yet diagnosis remains largely subjective and inaccessible. This project presents an end-to-end system that analyzes EEG (electroencephalogram) brainwave signals to classify depression, bridging neuroscience and mobile technology into a practical clinical tool.

Built independently as part of the HSIL Harvard Health Applications Hackathon 2026, this app demonstrates a complete pipeline from raw EEG signal processing to real-time mobile inference.

---

💡 Software Engineering Highlights

• Full-stack ownership — Sole developer across mobile, backend, and ML pipeline
• RESTful API design — Clean Flask endpoints decoupling frontend from ML logic
• Modular architecture — Signal processing, model inference, and app logic are fully separated concerns
• Cloud-native — Firebase handles auth, real-time sync, and storage with no custom infra
• Cross-platform — Single Flutter codebase targeting both Android and iOS

---

✨ Key Features

• 🔬 EEG Signal Processing — Applies Random Matrix Theory (RMT) denoising to remove noise artifacts from raw EEG recordings
• 🤖 ML Classification — Trained classifier on the MODMA dataset achieving 90%+ accuracy on held-out test data
• 📱 Cross-platform Mobile App — Flutter frontend with clean, intuitive UI for patients and clinicians
• ☁️ Cloud Backend — Python/Flask REST API hosted with Firebase for real-time data sync and authentication
• 🔐 Secure Auth — Firebase Authentication for user management

---

🏗️ Architecture

UI Screens
↓
State Management (Provider / Riverpod / Bloc)
↓
Services Layer
├─ ai_service.dart          → Python/Flask ML API calls
├─ database_service.dart    → Firestore real-time DB
└─ notification_service.dart

↓
Models
├─ doctor.dart
├─ patient.dart
└─ case.dart

↓
Cloud Database & Real-time Streams (Firebase)
↓
Secure Storage & Encryption

📁 Folder Structure

lib/
├─ main.dart
├─ screens/
│   ├─ splash_screen.dart
│   ├─ login_screen.dart
│   ├─ main_page.dart
│   ├─ new_cases_screen.dart
│   ├─ patient_history_screen.dart
│   ├─ chat_screen.dart
│   ├─ ai_recommendations_screen.dart
│   ├─ analytics_dashboard_screen.dart
│   └─ settings_screen.dart
├─ widgets/
│   ├─ case_card.dart
│   ├─ ai_card.dart
│   ├─ chat_bubble.dart
│   └─ bottom_nav_bar.dart
├─ models/
│   ├─ doctor.dart
│   ├─ patient.dart
│   └─ case.dart
├─ providers/
│   ├─ auth_provider.dart
│   ├─ chat_provider.dart
│   ├─ ai_provider.dart
│   └─ case_provider.dart
├─ services/
│   ├─ ai_service.dart
│   ├─ database_service.dart
│   └─ notification_service.dart
└─ utils/
    ├─ constants.dart
    └─ theme.dart

---

🧰 Tech Stack

Layer	Technology
Mobile Frontend	Flutter (Dart)
Backend API	Python, Flask
ML & Signal Processing	scikit-learn, NumPy, SciPy
EEG Denoising	Random Matrix Theory (RMT)
Database & Auth	Firebase Firestore, Firebase Auth
Dataset	MODMA (Multi-modal Open Dataset for Mental-disorder Analysis)

---

📊 Dataset

The model is trained on the MODMA Dataset — a publicly available, research-grade EEG dataset specifically curated for mental disorder analysis, including depression classification.

• EEG recordings from clinically diagnosed depressed patients and healthy controls
• Preprocessed using RMT-based denoising to isolate genuine neural signal from noise

---

🚀 Getting Started

Prerequisites

• Flutter SDK ≥ 3.0
• Python ≥ 3.9
• Firebase project with Firestore & Authentication enabled

Backend Setup

# Clone the repository
git clone https://github.com/Janamostafa-star/EEG-DEPRESSION-DETECT.git
cd EEG-DEPRESSION-DETECT

# Create virtual environment
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run the Flask server
python app.py

Flutter Setup

# Navigate to Flutter app directory
cd flutter_app

# Install dependencies
flutter pub get

# Run on device/emulator
flutter run

⚠️ Add your own google-services.json (Android) and GoogleService-Info.plist (iOS) from your Firebase console. These are excluded from the repo for security.

---

📁 Project Structure

EEG-DEPRESSION-DETECT/
├── flutter_app/          # Flutter mobile application
│   ├── lib/
│   └── pubspec.yaml
├── backend/              # Python Flask API
│   ├── app.py
│   ├── model/            # Trained ML model
│   └── requirements.txt
├── ml/                   # ML training & preprocessing scripts
│   ├── preprocess.py     # RMT denoising pipeline
│   └── train.py          # Model training
└── README.md

---

🔭 Future Work

• Improve model accuracy with deeper neural architectures (CNN/LSTM on raw EEG)
• Add real-time EEG streaming via Bluetooth headset integration
• Expand dataset with local Egyptian patient cohort (Egypt Vision 2030 alignment)
• Pursue regulatory pathway for clinical pilot deployment

---

👩‍💻 Author

Jana Mostafa Biomedical Engineering Student — AUC Cairo Built for the HSIL Harvard Health Applications Hackathon 2026

---

📄 License

This project is intended for academic and research purposes only. Not for clinical use without regulatory approval.