Healthcare

Microsoft diagnostic AI posts higher benchmark accuracy than doctors

Microsoft’s diagnostic AI reportedly outperformed generalist doctors on selected controlled case benchmarks, while questions remain about real-world clinical performance.

A report on Microsoft’s AI Diagnostic Orchestrator says the system was evaluated against medical case reports and emergency-department cases. The reported assessment used 304 cases from the New England Journal of Medicine and 325 emergency cases from the MIMIC-IV-ED database. In the controlled benchmark, the system achieved reported accuracy of 83.2% on the medical cases and 80.3% on the emergency cases. The comparison figures for generalist doctors were 36.1% and 46.8%, respectively. The system coordinates AI capabilities to work through diagnostic cases, making it an example of an AI approach aimed at clinical reasoning rather than a single image-recognition task. The report also stresses that the research took place in a controlled benchmark setting. The results therefore do not establish how the system would perform in everyday healthcare, where information may be incomplete and workflow, accountability and patient interaction matter. The development is significant because it highlights both the potential of advanced diagnostic systems and the need for real-world validation before clinical adoption.

Original source: bdnews24.com

Related RAQI solutions

In-depth analysis on RAQI Insight

Work with RAQI