Research
We ground AI in the structure that produces disease rather than in surface correlation — building systems that grasp not just that a patient is deteriorating, but why. The throughline is non-invasive sensing and structurally grounded inference applied to medicine.
Read our definition of computational medicineFocus Areas
Computational Medicine
Recovering clinically meaningful quantities — intraocular pressure, blood pressure, blood glucose — from signals that consumer hardware can capture.
Neuro-Symbolic AI & Active Inference
Agents that treat safety as a structural invariant, combining symbolic constraints with differentiable dynamics for principled, refusable decisions.
Physics-Constrained Modeling
Physically grounded, mechanistic models fused with learned components, so that data absorbs individual variation without abandoning physical structure.
Digital Health
Turning research into everyday tools — ambient, low-friction interfaces that bring screening and monitoring out of the clinic and into daily life.
Selected Publications & Presentations
- [1]From Black-Box to Glass-Box: Knowledge-Constrained Neuro-Symbolic Acute Pharyngitis Triage under Data Scarcity
Research Square (preprint), 2026. Oral presentation at ICRLSH 2026, Singapore.
Under review · Scientific Reports - [2]
DElenchus: Dynamic, Differential, Deep Active Inference with Hierarchical Markov Blankets for Causal Discovery
2026
Under review - [3]
Transfinite Anatomy Theory: Selective Hierarchical Boundaries for Disease Modeling
2026
Under review