Neurotrax
Neurotrax is an agentic AI system designed to make telehealth encounters a richer source of structured information about neurological function.
Core trick
Independent speech and facial analysis agents measure bounded features from telehealth video and abstain when signal quality is poor.
Description
Neurotrax is an agentic AI system designed to make telehealth encounters a richer source of structured information about neurological function. This creates a team of ambient agents that can assess and trend neurodegenerative disease progression in a standardized, quantitative way. Instead of treating video visits only as conversations, it coordinates independent speech and facial analysis agents that identify technically usable moments, measure bounded audiovisual features, abstain when signal quality is poor, and assemble the resulting evidence for clinician review.
README excerpt
# PhenoMetrix > Nonclinical research prototype. Not a medical device. No validated diagnostic, > treatment, emergency-detection, or disease-severity claims. Local development > uses synthetic participants; real-patient deployment requires separate > institutional authorization and integration work. PhenoMetrix derives quality-aware face and voice measurements from ordinary telehealth conversation. Its three capabilities are **Ambient Capture**, **Personal Trajectory**, and **Clinician Evidence Card**. The default integration direction is an analysis branch inside the patient's existing telehealth client, imported treatment context, and evidence available without interrupting care. The repository implements that integration substrate alongside the original local demo. It does not include a provisioned Zoom installation or a connected clinical EHR. See the [integration runbook](docs/encounter-integration.md), [architecture](docs/architecture.md), [development baseline](docs/development-baseline.md), and [validation boundaries](docs/validation.md). The [platform vision](docs/telehealth-platform-vision.md) describes the broader direction. ## Implemented encounter path ```text existing clinic identity, encounter binding, and modality-specific consent → authorized branch on the host's patient media tracks → local face/voice workers and bounded transient primitives → up to five-minute derived observations throughout the encounter → scoped API + append-only PostgreSQL h