Early disease detection and five-level triage that surface urgent cases and rank differentials for the clinician.
Conformal prediction that gives a statistical coverage guarantee honest confidence, not a black box.
Natural-language processing for medical records turning free text into structured, decision-ready data.
De-identification and split-identity architectures so models learn without ever seeing who the patient is.
Forecasting patient outcomes and operational risk to help teams intervene earlier and allocate better.
Designing decision support that keeps the expert in control and improves from their feedback.
Our research aims to revolutionize healthcare through AI improving patient care, streamlining processes, and enhancing medical decision-making. We hold ourselves to methods that are not only state-of-the-art, but safe, transparent, and fair.
Working on a clinical dataset or a research partnership? We're always interested in the hard problems.
Reach out