RESEARCH / OUR DIRECTION
Explore the decision.
Not just the answer.
Medicine unfolds over time. We want to understand how people and AI reason through it.
An early research direction. The questions below describe our ambitions, not established clinical findings.
THE CONNECTION
Clinical context.
Simulated possibilities.
Clinic software helps us understand how work moves around a patient. Simulation lets us explore possible decisions and their consequences in a constructed environment.
We see these as connected foundations for medical AI research: context that makes an action meaningful, and environments where proposed actions can be examined before real-world use.
OPEN QUESTIONS
What we want
to understand.
01Patient trajectoriesHow does a decision change what happens next?
We are exploring how simulation can represent a patient’s changing state, the timing of interventions and alternative courses of care. A useful simulation must make its assumptions and limitations visible.
02Medical AI evaluationCan an agent act appropriately, not just answer correctly?
Our research direction includes testing sequences of actions: what an agent notices, what it proposes, what it is permitted to do, and when it asks for help. Simulation is a setting for investigation, not proof that an AI system is safe in clinical use.
03Learning through decisionsWhat helps a learner develop better clinical reasoning?
We are interested in how repeated practice, feedback and alternative patient trajectories support learning, and how any benefit would need to be evaluated beyond the simulation itself.
04AI inside clinic workflowsWhere does assistance help, and where must a person stay in control?
Real clinic work gives us concrete questions about context, permissions, handoffs and review. We use those questions to shape tools that keep consequential decisions attributable to an accountable person.
A NECESSARY DISTINCTION
A simulated outcome
is not a clinical result.
Simulation can expose failure modes and help formulate better questions. It cannot replace clinical evidence, expert review or evaluation in the setting where a system will actually be used.
Work on these
questions with us.
We would like to hear from clinicians, educators, engineers and researchers with a concrete problem to investigate.
Discuss a collaboration