About
Building AI clinicians can trust
Research mission
Research
Research areas
Four pillars, one question: what does it take for a clinician to justifiably trust a model's output on their patient? Explore each area in depth →
Publications
Latest work
Currently under review. See all 14 publications →
News
Recent updates
CARMA AI Lab
Clinical AI for Reliable Medical Analytics
We audit models the way clinicians audit diagnoses — stress-testing them under distribution shift, explaining their reasoning, and quantifying when they should not be trusted.