Microsoft AI diagnostic tool's results and the limits of its comparison
What happened
The Verge reported on October 23 that a Microsoft study in Science says Diagnostic Orchestrator beat doctors on historical cases, with accuracy up to four times that of physicians. TechCrunch then reported that MAI-DxO scored 83% on simulated complex cases against 36% for participating general practitioners and general internists. The earlier report cited up to four times a doctor's accuracy; the later one gives 83% and 36%, and the two describe different test settings and measures. The later report also says the system coordinates multiple AI agents to take histories, order tests and discuss diagnoses, but the control group did not include the specialists or care teams who normally handle such cases, and its gain over the base model did not appear in emergency cases.
Written by AI from the coverage · updated 13 minutes ago
Coverage
Follow the reports to see the story from different sides.
- TechCrunch · AIMicrosoft’s Mustafa Suleyman says its AI outperforms doctors on hard cases as he chases ‘medical superintelligence’
Microsoft 在 Science 发表的研究显示，MAI-DxO 在复杂病例模拟中的诊断准确率为 83%，参与比较的全科及普通内科医生为 36%。该系统协调多个 AI 智能体问诊、安排检查并讨论诊断，但对照组未包括通常处理此类病例的专科医生或医生团队，其对基础模型的增益在急诊病例中也未显现。
- The Verge · AIMicrosoft says its AI tool is better than doctors at diagnosing conditions
Microsoft发表于Science的研究称，其Diagnostic Orchestrator在历史病例测试中的诊断准确率高于医生，最高可达医生的4倍。
Heat over time
Not enough continuous observations to draw a trend yet.