Virtual OSCE Examiner
A four-part simulated OSCE station: take a history from an AI patient, interpret investigation results, state your differential, and describe your management — then get a rubric-based examiner debrief covering all four tasks.
Press “Begin station” to start the ten-minute history (Phase 1).
You are the doctor. Ask one question at a time, in plain language. Enter to send · Shift + Enter for a new line.
A teaching demonstrator, not an assessment.
The point is deliberate practice with honest framing: the patient will not hand you the diagnosis during the history, and the debrief marks coverage — not whether you were right, and not a real OSCE score.
Coverage, not a mark
The checklist scores whether you asked about each rubric area and covered the expected content in each phase, graded by a language model. That is not a calibrated mark, a pass/fail, or a measure of clinical safety.
One scripted case
A single 20-year-old woman with appendicitis (Sophie Bungard). It is not a question bank, not interchangeable between sessions, and cannot substitute for varied clinical exposure.
The patient is a model
It follows a fixed case sheet and is told never to volunteer a diagnosis during history, but it can still be inconsistent or err. It is no substitute for a trained simulated patient or bedside teaching.
Diagnosis confirmed later
The AI patient will not name the diagnosis. The examiner confirms it in Phase 4 (management) as part of the structured flow — this is intentional, not a shortcut.
Out of scope: physical examination · multiple stations · saved attempts · validated scoring · real clinical advice.