EEG-powered hallucination detection and reality grounding for schizophrenia patients.
When the EEG detects hallucination-associated brainwave patterns, a camera scans the environment, AI describes what's actually there in calming language, and text-to-speech speaks it aloud — helping the patient distinguish hallucination from reality.
| Layer | Tool |
|---|---|
| EEG Hardware | Emotiv EPOC X (14-channel) |
| EEG API | Emotiv Cortex 2 (WebSocket JSON-RPC) |
| Computer Vision | YOLOv8 Nano |
| AI Narration | Gemini 2.5 Flash |
| Text-to-Speech | ElevenLabs |
| Backend | FastAPI + WebSockets |
| Frontend | Next.js 14 + Tailwind + Recharts |
See docs/SETUP.md for full instructions.
# Backend
cd backend && pip install -r requirements.txt
uvicorn main:app --reload
# Frontend
cd frontend && npm install && npm run devSee docs/DEMO_SCRIPT.md.
Use the Demo Mode button on the dashboard to simulate escalating EEG metrics and trigger the full pipeline without requiring a real hallucination event.