Alibaba DAMO Academy's AI Model DAMO COCA Enables Non-Invasive Colorectal Cancer Detection, Published in Top International Journal

Alibaba DAMO Academy, together with Guangdong Provincial People's Hospital and other institutions, developed the colorectal cancer screening AI model DAMO COCA, enabling non-invasive "painless" screening without bowel preparation. The result was published in the top international oncology journal Annals of Oncology. The model accurately identified 5 missed colorectal cancer cases from over 27,000 plain CT images, with a sensitivity of 86.6% and specificity of 99.8%, and a misdiagnosis rate of only 0.2%.

Compared with 10 radiologists, the model's sensitivity was 20.4% higher, especially in easily missed areas. With AI assistance, doctors' sensitivity and specificity could increase by 14.5% and 3.1% respectively, effectively reducing missed diagnoses.

This is the third cancer screening AI model released by DAMO Academy after pancreatic and gastric cancer models, marking the official completion of its "plain CT + AI" multi-cancer screening technology route, which is expected to solve the problem of low acceptance of conventional screening.