A recent Science paper reports that Microsoft’s MAI‑DxO system correctly identified 83% of complex cases, compared with a 36% accuracy rate for generalist doctors.

Microsoft’s diagnostic AI, MAI‑DxO, has demonstrated a striking advantage over generalist physicians, correctly identifying 83% of complex medical cases in a new study, while doctors achieved just 36% accuracy.

Study Overview

The research, published in the journal Science, evaluated the AI system against a cohort of generalist physicians using a set of notoriously difficult diagnostic scenarios drawn from real‑world patient records.

Performance Metrics

MAI‑DxO’s 83% success rate reflects its ability to integrate vast medical literature, imaging data, and patient histories, whereas the physicians’ 36% accuracy highlights the challenges faced when confronting atypical presentations without specialist support.

  • AI correctly diagnosed 83% of cases
  • Physicians correctly diagnosed 36% of cases
  • Study included 200 anonymized patient records
  • Evaluation covered multiple specialties including oncology and neurology

Implications for Healthcare

If deployed responsibly, tools like MAI‑DxO could serve as decision‑support aides, helping clinicians navigate complex cases and potentially reducing diagnostic errors. For more on how AI is reshaping the sector, see the AI coverage on uxc.news and Microsoft’s profile on Momentum.

TechCrunch coverage of Microsoft AI outperforms doctors