ZeroMotion

ZeroMotion is a privacy-preserving Progressive Web App (PWA) that translates complex knee surgery recovery metrics into clear patient insights while proving physical compliance to doctors using Zero-Knowledge Proofs (ZKPs)—without exposing raw workout videos, survey responses, or medical identity.

Key Features

• Local-First Processing: KOOS questionnaires and motion checks ($92^\circ$ flex, $88\%$ EMG engagement) remain strictly on-device.
• Anonymous AI Translation: Stripped parameters pass to NVIDIA NIM (meta/llama-3.2-90b-vision-instruct) to convert clinical figures into plain-English updates.
• Client-Side ZK Proofs: Generates cryptographic proofs of recovery milestones locally using Compact smart contracts.
• Shielded Doctor Ledger: Clinicians view anonymous patient keys and ZK-Verified badges via a real-time Server-Sent Events (SSE) pub/sub stream, unlocking recovery plans instantly on approval.

ZK Math Constraints

• KOOS Score Bounds Check: Validates score $K$ without revealing individual survey answers: $$ 0 \le K \le 100 $$

• Sensor Commitment Hash: Authenticates sensor streams without exposing raw exercise metrics: $$ H = \text{SHA-256}(\text{flexion_angle} \parallel \text{emg_engagement}) $$

Technical Architecture

Roadmap

1. Compile koos_range and session_commitment circuits via compactc.
2. Replace local key witness engine with Midnight Lace Wallet integration.
3. Deploy state transitions to the Midnight local developer network.

Built With

• blockchain
• bun
• compact
• healthcare
• llama3.2
• local-first
• midnight-network
• mobile-first
• nvidia-nim
• privacy
• pwa
• react
• server-sent-events
• tailwindcss
• tanstack-router
• tanstack-start
• typescript
• web3
• zero-knowledge-proofs