AMIE

Towards Conversational Diagnostic Artificial Intelligence (Articulate Medical Intelligence Explorer)

Paziente / Popolazione Intervento / Esposizione Confronto Esito

In 159 OSCE-style text consultations with validated patient-actors, using case scenarios from providers in Canada, the UK and India, randomised in a crossover design between AMIE and 20 primary care physicians, AMIE, an LLM-based conversational diagnostic AI improved diagnostic accuracy and rated consultation quality compared with primary care physicians.

N
159
Disegno
Randomised, double-blind crossover OSCE-style study, scenarios randomised, Canada/UK/India
Esito
Diagnostic accuracy and performance on clinically meaningful axes (history-taking, diagnosis, management, communication, empathy) rated by specialist physicians and patient-actors
Rilevanza
2
RisultatoAMIE showed greater diagnostic accuracy and was superior on 30 of 32 axes rated by specialist physicians and 25 of 26 axes rated by patient-actors; the abstract reports no effect sizes.
Tu T, et al. Nature. 2025;642(8067):442-450. 10.1038/s41586-025-08866-7
Discussione e critica

The first randomised, blinded head-to-head of a conversational AI against physicians across history-taking, diagnosis and empathy — a milestone, and the AMIE system was later tested in specialist care (O'Sullivan LLM (cardiology)). Heavily caveated by its own authors: synchronous text chat is unfamiliar in clinical practice, the patients were actors, and nothing was tested in real-world care.