MedGPT outperforms doctors in 88% of lung-cancer case comparisons in Chinese multicenter study
Reported by Dr. Kai Wei, Frontier AI Desk
- MedGPT scored higher than doctors in 88% of head-to-head evaluations on 40 complex cases.
- Average physician score rose from 53.40 to 66.27 when assisted by MedGPT, a 24.1% relative gain.
- MedGPT achieved a 96.22 score versus 71.38 for GPT-5.4, 61.81 for Claude Opus 4.7 and 54.62 for Gemini 3.1 Pro.
Future Doctor’s MedGPT model beat unassisted physicians in 660 of 750 lung-cancer treatment strategy comparisons.
«Future Doctor Says MedGPT Outscored Doctors in 88% of Comparisons on Complex Lung Cancer Cases»
Summary and commentary by Aura-V. Facts, figures and opinions belong to the original.
Future Doctor released results of a multicenter evaluation in which its MedGPT model produced higher-scoring treatment strategies than unassisted doctors in 660 of 750 comparisons (88%) on 40 complex lung-cancer cases, according to a report on Pandaily. The study involved 250 doctors from 98 centers across 27 provincial-level regions, 63.6% of whom were chief or associate chief physicians. Each clinician generated three strategies per case under three conditions: unassisted, with literature search, and with MedGPT, yielding 2,250 strategies.
Twenty multidisciplinary experts created 240 case-specific scoring rules, and three AI judges—GPT-5.5, Claude Opus 4.7 and Gemini 3.1 Pro—scored the strategies blind to their source. MedGPT’s average score was 96.22, far above the 71.38 achieved by GPT-5.4 and well above the scores of the other baseline models.
Our read: The self-reported figures suggest MedGPT can substantially augment clinician decision-making in ambiguous lung-cancer cases, but the reliance on AI judges from the same model families and the absence of patient-outcome data limit the claim’s clinical robustness.
Interested in implementing this technology inside your company?
Our studio designs custom agent meshes, data pipelines, and outbound GTM systems for forward-looking enterprises.