Editor’s summary : AI applications for clinical medicine continue to advance but often rely on written vignettes. Unfortunately, these are not a realistic representation of clinical medicine, where patients do not come in with a tidy write-up of key symptoms and pertinent laboratory findings. To address this issue, Nori et al. made their AI system start from the patient’s chief complaint and decide what follow-up questions to ask and what testing to order to collect the information needed for a diagnosis. In multiple test settings, this system achieved higher accuracy than human responders and at a lower total cost, demonstrating its potential as a tool to assist physicians. —Yevgeniya Nusinovich
Abstract
We explored the capabilities of frontier language models in sequential diagnosis, a diagnostic paradigm that captures the iterative nature of clinical reasoning. Given a brief clinical presentation, clinicians selectively acquired information to reach a diagnosis, enabling assessment of diagnostic accuracy and workup cost. We developed an interactive benchmark and an artificial intelligence (AI) system that simulates a panel of physicians, continually revising the differential diagnosis and selecting cost-effective tests. Using this benchmark, the system achieved 83.2% accuracy on New England Journal of Medicine clinicopathological conference cases and 80.3% on MIMIC-IV-ED emergency department admissions, outperforming generalist physicians (36.1 and 46.8%, respectively). Although these findings were obtained in a controlled benchmark setting, they highlight the promise of language models for supporting more accurate, lower-cost diagnoses.
# sources
Posté par yinqi . Évalué à -1 (+0/-2).
Sequential diagnosis with language models
Votre médecin devra-t-il bientôt utiliser une IA pour bien vous soigner ?
Diagnostic médical : quand l’intelligence artificielle imite la démarche, étape par étape, des médecins
Envoyer un commentaire
Suivre le flux des commentaires
Note : les commentaires appartiennent à celles et ceux qui les ont postés. Nous n’en sommes pas responsables.