Microsoft’s MAI-DxO AI outperforms doctors on complex cases but costs more to run
Microsoft’s MAI-DxO diagnostic system scored 83.2% accuracy on complex medical cases, compared with 36.1% for generalist physicians using reference materials, according to research published in Science on October 22. The system also outperformed doctors on emergency department cases (80.3% vs. 46.8%). The results, based on written case records—not live patients—highlight potential but do not…
Key points
- Microsoft’s MAI-DxO AI scored **83.2%** accuracy on complex medical cases, outperforming doctors at **36.1%** on the same benchmarks
- Orchestrating AI models like GPT-5 improved accuracy from **78.9%** to **83.2%**, but the system asked **25.5 questions per case** vs. **5.6 for doctors**
- Cost estimates range from **$1,303 to $5,431 per case**, higher than physicians’ **$4,725**, though cheaper configurations retained **80%+ accuracy**
The story so far
2 episodes →- Microsoft’s MAI-DxO AI outperforms doctors on complex cases but costs more to runthis story
Microsoft's AI Diagnoses Complex Diseases Better Than Doctors: But It Costs More to Run
ibtimes.sg · 26 October 2026
Loading the full article…
This text was published by ibtimes.sg and written by https://www.ibtimes.sg/reporters/dheemanth-kashyap. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Enterprise & Industry
All →- Oracle adds six AI agents to speed Oracle Fusion Cloud deployments · 6 src
- The RealReal tightens bot defenses after AI agent exploits reserve system · 1 src
- AWS builds agentic retrieval on managed Bedrock knowledge base with CDK · 1 src
- Oracle NetSuite adds AI agents and skills to automate business workflows · 7 src
- OpenAI reportedly bans user for reverse-engineering TV · 1 src
Comments
via GitHub Discussions