Alibaba Damo Academy’s AI achieves “no‑prep” colon‑cancer screening on plain CT, reportedly lands in top oncology journal

Breakthrough: opportunistic, patient‑“no‑feeling” screening

Alibaba Damo Academy (阿里巴巴达摩院), together with Guangdong Provincial People's Hospital (广东省人民医院) and other institutions, has developed an AI model called DAMO COCA that reportedly performs opportunistic colon‑cancer screening on plain (non‑contrast) CT scans without bowel preparation. It has been reported that the work has been published in a leading international oncology journal, marking what Alibaba describes as the first global proposal of a truly “no‑prep” colon screening pathway. Could routine CTs taken for other reasons become a low‑burden way to catch colon cancers earlier?

Study and performance

According to the report, researchers scanned a dataset of about 27,000 plain CT exams and used DAMO COCA to flag five previously missed colon cancers. The model’s sensitivity and specificity were reported at 86.6% and 99.8%, respectively. Reportedly this is Damo’s third cancer screening AI after models for pancreatic and gastric cancers, and Alibaba says the result shows the “plain CT + AI” multi‑cancer screening approach is technically feasible.

Implications, limits and geopolitical context

If validated externally, opportunistic “no‑prep” screening could expand early detection in settings where CT imaging is already common — a practical advantage in China’s large hospital system. But the absolute number of identified cancers in the study was small and external, prospective validation and regulatory review are needed before clinical rollout. It has been reported that commercial and international deployment could face additional hurdles: data‑governance scrutiny, national medical‑device approvals, and geopolitical constraints on cross‑border health technology adoption. For Western clinicians and regulators, the question will be whether performance holds up in diverse populations and in prospective workflows.