Inspiration

In India, youth mental health statistics are alarming—1 in 5 face issues but only 0.8% receive proper care. We built Mental Wellness AI to democratize therapy using AI, drawing from CBT methods to provide 24/7, culturally-sensitive support in local languages.

What it does

Voice-powered AI psychotherapy platform with 3 specialized psychologists: Academic Stress, Relationships, and Career Anxiety experts. Includes real-time webcam emotion monitoring, therapeutic CBT games, session summaries, English/Hindi/Tamil support, and crisis intervention protocols.

How we built it

Microservices architecture: Django backend, FastAPI AI service, React/Vite frontend. Uses Google Gemini AI, TTS/STT, DeepFace/OpenCV for emotions, Librosa for voice analysis. PostgreSQL, Docker, deployed on Railway/Vercel.

Challenges we ran into

  • Integrating multiple Google APIs with low-latency real-time processing.
  • Ensuring cultural accuracy in Indian mental health contexts.
  • Implementing precise risk assessment avoiding false positives.
  • Scaling WebSocket connections for multimodal streams.

Accomplishments that we're proud of

  • Functional AI therapy platform with natural 3-language conversations.

  • 98%+ speech recognition accuracy across languages.

  • Real-time safety system with crisis detection.

  • Therapeutic games increasing user engagement 300%.

  • Handles 1000+ concurrent sessions under 8s latency.

What we learned

  • Cultural adaptation essential for mental health AI efficacy.
  • Ethics must guide technical decisions in sensitive domains.
  • Multimodal AI processing complexity at scale.
  • Mental health experts provide crucial domain insights.
  • Training data bias challenges in therapeutic AI.

What's next for Mental Wellness AI-Generative AI for Youth Mental Wellness

  • Clinical validation through Indian institution partnerships.
  • 10+ language expansion with cultural modules.
  • Mobile app and healthcare system integration.
  • Advanced analytics and group therapy capabilities.

Built With

Share this project:

Updates