Research across healthcare AI.

Clinically grounded AI systems that shift healthcare from reactive treatment to predictive medicine. We develop models that simulate physiological systems, forecast disease progression, and guide safer interventions.

Our work bridges machine learning with domain expertise in medicine, ensuring every model respects clinical constraints and integrates into care workflows. Predictive modeling paired with physiological simulation gives clinicians tools they can trust.

Safe deployment in clinical settings is non-negotiable. We design for interpretability, uncertainty quantification, and rigorous validation, so AI augments clinical judgment instead of replacing it.

Personalized wellness models grounded in behavior science and designed around the people they serve.

Health outcomes depend on people, not just predictions. We design AI for wellness and health coaching that is grounded in behavior science and shaped by the lived context of the people using it, accounting for motivation, culture, and trust so that guidance is not only accurate but genuinely actionable.

Trust has to be earned by the system, not assumed. We test our models with real users and real routines, refining how guidance is delivered so it fits naturally into daily life instead of adding another chore to track.

High-impact, sustainable healthcare models engineered for low-compute efficiency to cut energy costs.

Clinical impact should not require a data center. We build small, efficient models that deliver high performance at a fraction of the compute and energy cost, making trustworthy healthcare AI cheaper to run, easier to deploy at the point of care, and more sustainable at scale.

Efficiency is a design constraint from day one, not an afterthought. We measure energy and compute cost alongside accuracy, so a model only ships once it proves it can run where care actually happens, including low-resource settings.