Skip to content

About

Research platform for simulating clinical smart-glasses workflows, device contracts, capture, voice, and safe-stop states.

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

6 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Smart Glasses Clinical Platform

Clinical smart-glasses workflows are difficult to test when device behavior, capture, voice, and safe-stop states are not modeled together. Smart Glasses Clinical Platform is an Android-first research and prototyping workspace that simulates those contracts before hardware deployment.

The first hardware target is HeyCyan-compatible camera glasses. The architecture deliberately keeps the device driver behind a small capability API so the same clinical workflows can later run on MentraOS-compatible devices, Brilliant Frame/Halo, Vuzix, or proprietary hardware.

Start here

  1. Read docs/00-project-overview.md.

  2. Review the docs/research/open-source-repository-review.md and the HeyCyan docs/protocols/heycyan-compatibility.md.

  3. For surgeon-facing work, read docs/surgery/START_HERE.md.

  4. Run the device and surgical-session simulators:

    python3 -m unittest discover -s tests -v
    python3 tools/device_simulator.py --scenario happy-path
    python3 tools/device_gateway.py --port 8767
    python3 tools/surgical_session_simulator.py
    python3 tools/ot_runtime_simulator.py

    For the dependency-free synthetic JSON gateway, run python3 tools/device_gateway.py --port 8767. It provides GET /health, GET /api/state, GET /api/events, POST /api/command, and POST /api/reset on loopback only.

    Example command:

    curl -s http://127.0.0.1:8767/api/command \
      -H 'content-type: application/json' \
      -d '{"name":"device.connect","command_id":"connect-1"}'

    The gateway is synthetic and fail-closed: it does not connect to physical glasses, accept clinical navigation commands, or store patient data.

    For the multi-adapter frontend/backend integration lab:

    ./scripts/bootstrap.sh
    .venv/bin/python tools/integration_server.py

    Open http://127.0.0.1:8766. The simulator supports discovery, connection, camera/photo/video/audio/stream lifecycle, media metadata, browser-local camera/microphone capture, WebSocket events and push-to-talk commands. HeyCyan, Mentra and Brilliant profiles remain unavailable until their authorized external bridge URL is configured; see docs/architecture/device-integration-runtime.md.

  5. For a physical Android/HeyCyan test, follow apps/android-controller/README.md.

Open-source and research boundary

The original project code is available under Apache-2.0. See LICENSE, NOTICE, CONTRIBUTING.md, GOVERNANCE.md, and COMPLIANCE.md. Third-party SDKs, source archives, datasets, and models retain their own terms and require separate provenance review. A version or passing test suite does not imply clinical validation, regulatory clearance, or production readiness.

Repository map

apps/android-controller/       Android companion and HeyCyan adapter boundary
packages/device-contracts/     Vendor-neutral device events and commands
packages/device_runtime/       Multi-adapter runtime and external bridge clients
packages/surgical-workflows/   Non-clinical OR session and safety-state contracts
packages/media-runtime/         Bounded low-latency video and image-quality controls
packages/voice-runtime/         Narration and allowlisted voice-command contracts
services/clinical-gateway/     Future PHI-aware policy and integration service
services/ot-runtime/            Synthetic OT-safe media/voice session orchestrator
tools/                         Deterministic local device simulator
tools/integration_server.py    FastAPI multi-adapter backend and webapp server
tools/ot_runtime_simulator.py  End-to-end synthetic media/voice/OT smoke tool
webapp/                        Calm local simulator console and method page
tools/lidar_wound_depth/       Calibrated LiDAR surface-depth research tool
tools/lidar_wound_progress/    Standalone longitudinal LiDAR geometry reviewer
tools/lidar_wound_progress/webapp/  Local-first camera/depth review webapp
tools/lidar_wound_progress/local_service.py  Loopback SQLite summary service
tests/                         Protocol/state-machine tests
docs/                          Architecture, research, safety, product and test plans
third_party/                   License records and placeholders; no unlicensed SDK code

Current scope

  • simulated connection lifecycle (physical BLE discovery remains hardware-gated)
  • battery, version and media-count queries
  • photo, video and audio controls
  • media-count queries; transfer interruption and checksum handling remain planned
  • device simulator and event log
  • declared capability enforcement and safe-state rules in the simulator
  • idempotent command replay behavior in the simulator
  • capability-mismatch rejection before device actions
  • active-session safe stops for disconnect, critical battery, and capture-indicator loss
  • groundwork for clinical capture, telepresence and documentation workflows
  • vendor-neutral LiDAR surface-depth measurements with synthetic wound-surface fixtures
  • standalone longitudinal geometry signals with synthetic serial LiDAR captures
  • a runnable, synthetic surgical-observation session state machine
  • bounded low-latency frame buffering with keyframe-aware dropping
  • luma-based transport quality gates and adaptive media profiles
  • transcript/narration contracts and confirmation-gated voice commands
  • synthetic OT runtime with preflight, voice start/stop, live frame publishing, and disconnect-safe stopping

This repository is a research prototype. It must not be used for diagnosis, treatment selection, surgical navigation, or production handling of patient data.

The Android adapter sources define a licensed-SDK boundary but are not yet a buildable Android application. Physical-device behavior, permissions, thermal limits, media checksums, and the validation targets in docs/testing/device-validation-plan.md remain hardware-gated.

The media and voice runtimes are intentionally adapter-first. They are functional with synthetic frames and transcripts; real glasses streaming, continuous speech recognition, codec tuning, and operating-room deployment still require hardware, network, permission, safety, privacy, security, and human-factors validation.

About

Research platform for simulating clinical smart-glasses workflows, device contracts, capture, voice, and safe-stop states.

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages