We built Dx to give students a safe place to practice speaking, listening, and connecting with patients without the consequences of a real-world misdiagnosis. Each session begins with a simulated AI patient, whose symptoms and history provide clues that students must uncover through conversation.
Behind the experience, Gemini turns ranked disease and symptom data from Columbia University Biomedical Informatics into structured patient cases. ElevenLabs powers the voice conversation, while our Next.js interface and Python backend manage accounts, cases, simulation history, diagnoses, and feedback reports.
The result is a workflow that takes students from meeting a patient to submitting a diagnosis and reviewing which symptoms supported their reasoning or pointed elsewhere. An insights panel also helps them look more closely at symptoms they are less likely to diagnose accurately.
Check out our live website and see if you can diagnose the patient correctly! (Tip: when in doubt, it seemed to always be bronchitis...)