# Ontology-Physiology Lab

- **Event:** [Meta Global AI Developer Hackathon](https://cerebralvalley.ai/e/meta-ai-developer-hackathon)
- **When:** Oct 26, 12:00 PM – Nov 4, 9:00 PM (EST)
- **Where:** Online
- **Team:** [Sungyeob Lim](https://cerebralvalley.ai/u/cgcorporation)
- **GitHub:** https://github.com/Carraway-and-Gatsby/ontology-physiology-lab
- **Demo video:** https://youtu.be/JWoJ1BdVDyg
- **Gallery:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/670

Ontology-Physiology Lab turns a plain-language cardiovascular model hypothesis into a repeatable computational experiment with a report that can be checked against the actual simulation.

A researcher can ask, “In M1, halve LV peak elastance and compare cardiac output.” Meta Muse Spark maps the request to a bounded parameter catalogue, selects an intervention, calls the simulation tools, and chooses the runs and metrics to report. The deterministic engine supports a six-compartment circulation model and an eight-volume four-chamber model. Ontology annotations supply physical-property, anatomical and unit context for inspecting the parameter mapping.

Code builds the report from retained tool receipts, including values, units, parameter changes, warnings and evidence pointers. A separate verifier checks the report against those receipts, and every session records W3C PROV-O provenance. The public Lab lets judges replay recorded model calls, re-execute the engine in their own browser through Pyodide, and re-verify the report.

A recorded evaluation of Muse Spark passed all checks in 27 author-written cases: 13 synthetic experiments and 14 out-of-scope clinical requests. The clinical requests were stopped by the deterministic guard before model calls. All 13 experiment cases completed with warnings, including the short simulation duration and periodicity diagnostics. This is a small engineering evaluation, not evidence of clinical accuracy or superiority over other models.

The project is intended for researchers exploring and auditing small physiological models. All parameters are synthetic and exploratory; it is not clinically validated, patient-specific, or a medical-advice system. Ontology coverage remains incomplete, and hypothesis-to-parameter mappings require researcher review.

Interactive demo: https://medical-ontology-research.pages.dev/lab/

## More from Meta Global AI Developer Hackathon

- [Djinn Council](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/667)
- [Momar SOZO](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/668)
- [Cortex AI](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/669)
- [Maina Lewis](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/671)
- [Vijayanand Settin](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/672)
- [furniture-moving rehearsal](https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/673)

---

Markdown version of https://cerebralvalley.ai/e/meta-ai-developer-hackathon/hackathon/gallery/670. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt