Fizzio is a digital physical therapy platform that bridges the gap between clinical guidance and at-home exercise. By combining on-device computer vision with agentic AI, Fizzio provides real-time posture correction and dynamic workout scheduling that adapts to your recovery progress.

Fizzio uses on-device computer vision to track 33 3D body landmarks. The app calculates precise joint angles (e.g., knee extension, hip flexion) to ensure exercises are performed safely.

• Live Coaching: Receive immediate audio and visual cues to correct your form.

Unlike static exercise templates, Fizzio employs an agentic AI approach to rehabilitation.

• Personalized Generation: Initial plans are generated based on specific pain descriptions or clinical diagnoses.
• Autonomous Adjustments: The AI agent analyzes your performance (reps completed, hold times) and subjective feedback (pain scales and difficulty) to modify the intensity, volume, or exercise selection for subsequent sessions.
• Structured Recovery: Ensures you are challenged enough to progress without risking re-injury.

Monitor your recovery journey through detailed analytics.

• Mobility Trends: Track improvements in joint range of motion over time.
• Consistency Logs: View your adherence to the prescribed rehabilitation schedule.
• Pain Mapping: Correlate exercise consistency with reported pain levels to see tangible recovery results.

• Frontend: React Native (Expo) with TypeScript
• Backend: Python (FastAPI)
• Computer Vision: MediaPipe Pose
• Database & Auth: Supabase (PostgreSQL)
• State Management: Zustand
• AI Orchestration: Gemini Pro / OpenAI (via Structured Outputs)

• Node.js (v18+)
• Python 3.10+
• Expo Go app on your mobile device (for testing camera features)
• Supabase Account

1. Navigate to the backend directory:

   cd backend

2. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Configure environment variables in a .env file:

   SUPABASE_URL=your_project_url
   SUPABASE_KEY=your_service_role_key
   OPENAI_API_KEY=your_key  # or GEMINI_API_KEY

5. Start the server:

   uvicorn main:app --reload

1. Navigate to the frontend directory:

   cd frontend/myphyspal

2. Install dependencies:

   npm install

3. Configure your API endpoint: Create a .env file in the myphyspal directory and add your backend URL and Supabase credentials.
4. Start the Expo development server:

   npx expo start

├── backend/
│   ├── main.py              # FastAPI entry point
│   ├── supabase_client.py   # Database integration
│   └── requirements.txt     # Python dependencies
└── frontend/myphyspal/
    ├── app/                 # Expo Router file-based navigation
    ├── components/          # Reusable UI components (Camera, UI, Survey)
    ├── hooks/               # Custom React hooks (Theming, Auth)
    ├── lib/                 # Shared logic (API, Supabase, Profiles)
    └── store/               # Zustand state management

Fizzio operates on a feedback loop that treats the workout plan as a living document. After every session, the backend triggers an analysis step where the AI agent evaluates the session's "Success Score." If the user reports high pain but high completion, the agent may down-regulate intensity. Conversely, if completion is high with low RPE, the agent autonomously increases the challenge for the next session, ensuring the plan remains optimal for recovery.