HARVARD MEDICAL SCHOOL · BIOMEDICAL INFORMATICS

Medical AI for
complex clinical work.

We build AI that reasons from evidence, test it before we trust it,
and teach it to assist.

Conceptual illustration: patient history, lab results, and vital signs connect to a clinical decision and a clinician–patient conversation. — EvidenceDecision

Reason from evidence.

We build AI that reads scans and records the way specialists do, ties each answer to its evidence, and finds risk no one asked it to look for.

GOALAI that predicts each patient's future from every scan and every record, so clinicians can act before disease arrives.

Conceptual illustration: an AI agent acts in a simulated hospital and receives new observations, forming a continuing feedback loop. — AI agentSimulated care

Test before we trust.

We test AI the way it will be used: talking with patients, moving between hospitals, working beside clinicians, and acting in simulated care.

GOALA simulator of an entire hospital, with every department, bed, and handoff, so AI can be tested on a whole hospital's work before it meets a patient.

Conceptual illustration: an expert demonstrates a movement over a practice pad, and a supervised robot learns a matching assistive movement. — DemonstrationAssistance

Learn to assist.

We teach robots to assist clinicians with their hands, in the operating room and at the bedside, with the clinician in control.

GOALA portable surgical assistant that rolls up to any operating table, and robotic imaging that brings expert scans to every patient, wherever they are.

Shown at 2× speed

LEARN TO ASSIST

A robot that follows the suture for the surgeon.

Closing a wound pulls a long tail of thread across the field, and an assistant usually holds it out of the way. Trained on under an hour of demonstrations, a low-cost robot arm followed 23 of 25 stitches alongside a surgeon.

Imitation Learning for Robot Assistance in Open Surgery MLHC · 2026

SELECTED RESEARCH