SafeSpace is a Flutter-based mobile application focused on mental health, counseling, and safe community interactions.
The app provides users with a supportive platform to connect with counselors, access resources, and engage in anonymous group discussions.

• 🔐 Authentication – Secure sign-up and sign-in with Supabase
• 🎧 Entertainment & Relaxation – Meditation, music, and interactive exercises
• 🤖 AI Chatbot – Intelligent mental health assistant powered by Groq LLM with real-time responses
• 🧠 Mental State Analysis – Automatic analysis of user messages to predict mental health states
• 💡 AI Suggestions – Personalized wellness recommendations based on dominant mental state
• 👨‍⚕️ Doctor Recommendations – Smart matching with mental health professionals based on mental state
• 🎬 Entertainment Recommendations – Curated content matching user's emotional needs
• 🆘 Crisis Detection – Immediate crisis response with emergency hotline numbers
• 👨‍⚕️ Counselor Channeling – Book appointments and connect via Google Meet (Future development)
• 📝 Forum / Blog – Share thoughts, stories, and mental health tips (Future development)
• 💬 Anonymous Group Chat – Engage in peer-to-peer conversations safely (Future development)

• Frontend: Flutter (Dart)
• Backend: FastAPI (Python) with Groq LLM integration
• Database: Supabase (PostgreSQL) with real-time subscriptions
• Authentication: Supabase Auth
• Storage: Supabase Storage
• AI/ML:
  • Groq API (LLaMA 3.1 for conversational AI)
  • Mental state prediction with heuristic fallback
  • Transformers & PyTorch for NLP
• Real-time: Supabase WebSocket subscriptions
• Integrations: Google Meet API (Future)

• Install Flutter (>=3.27.3)
• Install Dart
• Setup Supabase Project
• Configure your API Keys and Environment Variables

# Clone the repository
git clone https://github.com/DhanukaRathnayaka/Final_Year_Project.git

# Navigate to project directory
cd Final_Year_Project

# Get Flutter dependencies
flutter pub get

# Set up your Supabase configuration
# Create a new file lib/supabase_config.dart with your Supabase credentials:
# ```dart
# const supabaseUrl = 'YOUR_SUPABASE_URL';
# const supabaseAnonKey = 'YOUR_SUPABASE_ANON_KEY';
# ```

# Run the Flutter app
flutter run

If you don't want to run the backend locally and instead use the deployed API (for example, https://backend-t3db.onrender.com), the Flutter app already supports a compile-time switch.

Run with the deployed backend by passing a dart-define flag:

# Web
flutter run -d chrome --dart-define=USE_DEPLOYED_BACKEND=true

# Android or iOS
flutter run --dart-define=USE_DEPLOYED_BACKEND=true

This instructs the app to use the configured deployed URL. If you want to hardcode the deployed URL without the flag, see lib/config.dart and set useDeployedBackend to true.

Note: The deployed backend has CORS enabled (see lib/Backend/app.py), so there should be no origin issues when accessing it from web or mobile clients. If you experience any CORS errors, open the deployed API logs and ensure allow_origins is configured correctly.

The backend server is located in the lib/Backend directory and consists of:

• FastAPI server for the main API
• Mental Health Chatbot with sentiment analysis
• Supabase integration for data storage

# Navigate to backend directory
cd lib/Backend

# Create and activate virtual environment
python -m venv venv
.\venv\Scripts\activate  # On Windows
# source venv/bin/activate  # On Linux/Mac

# Install dependencies
pip install -r requirements.txt

# Install additional AI-related packages
pip install transformers torch numpy fastapi uvicorn python-dotenv supabase

Create a .env file in the lib/Backend directory:

# Supabase Configuration
SUPABASE_URL=your_supabase_url
SUPABASE_KEY=your_supabase_service_key

# Groq API Configuration (Required for AI Chatbot)
GROQ_API_KEY=your_groq_api_key

# Optional - OpenAI API
OPENAI_API_KEY=your_openai_api_key

# Debug Mode
DEBUG=false

Note: To get a Groq API key, visit Groq Console

Create the following tables in your Supabase database:

Core Tables:

• conversations – Store chat conversations and sessions
• messages – Store individual messages in conversations
• users – User profiles and metadata

Mental Health Analysis:

• mental_state_reports – Store mental state analysis results with dominant state and confidence

Recommendations:

• doctors – Mental health professionals database
• entertainments – Entertainment content (music, videos, exercises)
• suggestions – Daily wellness suggestions database
• recommended_doctor – User-doctor assignments
• recommended_entertainments – User entertainment recommendations
• recommended_suggestions – User suggestion recommendations

For detailed schema and RLS policies, see the documentation files in the project root.

# From the lib/Backend directory
uvicorn app:app --reload --port 8000

The backend API will be available at: 👉 http://localhost:8000

Interactive API docs will be available at: 👉 http://localhost:8000/docs

Deployed API docs (Render): 👉 https://backend-t3db.onrender.com/docs#/

• POST /api/chat – Send message to AI chatbot

• POST /api/predict-mental-state – Predict mental state from text

• GET /ai-suggestions/suggestions/{user_id} – Get personalized AI suggestions
• GET /recommend_entertainment/api/suggestions/{user_id} – Get entertainment recommendations
• POST /recommend – Get doctor recommendations
• GET /api/suggestions/{user_id} – Get combined suggestions

For detailed API documentation, visit the API Documentation.

Comprehensive documentation is available:

• Backend & Chatbot Flow – Complete system architecture and flows
• Architecture Diagrams – Visual representations of all processes
• Implementation Guide – Setup, configuration, and troubleshooting

The SafeSpace system works as follows:

1. User Authentication → User logs in via Supabase Auth
2. Chat Initiation → User starts conversation with AI chatbot
3. Real-time Messaging → Messages sync via Supabase real-time subscriptions
4. AI Processing → Groq LLM generates compassionate responses
5. Mental State Analysis → After 5+ messages, system analyzes mental state
6. Personalized Recommendations → Based on dominant mental state:
   • AI-generated wellness suggestions
   • Entertainment content matching mood
   • Doctor matching by specialization
7. Crisis Detection → If crisis keywords detected, immediate hotline numbers provided

• Kavindu Dedunupitiya – Project Lead and UX UI Designer ( 22UG1-0812 )
• Dhanuka Rathnayaka – Fullstack Developer ( 22UG1-0828 )
• Gayanga Bandara – Fullstack Developer (22UG1-0285)