Alibaba Open-Sources Medical AI: AUC 0.913 on 40,000 Scans
Alibaba has open-sourced Damo Radar, a medical AI model for analyzing abdominal CT scans. Tested on 40,000 real exams, it covers around 150 pathologies including cancer, with a mean AUC of 0.913.
Alibaba has open-sourced Damo Radar, a medical model that analyzes contrast-enhanced CT scans and detects around 150 abdominal pathologies, including cancer. According to a paper in Science, the model was tested on 40,000 real exams. The mean AUC across 146 clinical findings was 0.913 out of 1.0. The model outperformed most radiologists.
This cannot replace a doctor without clinical validation. That is obvious, and the vendor does not claim otherwise. But open weights for a medical model at this level are already a working foundation. There is no need to build from scratch something that can detect cancer at the level of most specialists.
Source: www.scmp.com
Free course
Stop reading about AI — start building with it
The free Claude Code course: your first site, tool or game — no coding. No upsells, no cross-sells — nothing to buy here.
Start free →
Author
Evgenii Arsentev
PhD · Chief Executive Officer, digital health
Articles · Latest articles