AI catches more polyps in the real world. A large quality improvement trial across Veterans Health Administration facilities found that AI-based computer-aided detection (CADe) devices during colonoscopy raised adenoma detection rates by 22% compared to standard care. Notably, the benefit held across all endoscopists — even high performers — suggesting CADe could be a meaningful upgrade for colonoscopy quality nationwide.
AI catches more polyps in the real world
Randomized trials have long shown that AI-assisted colonoscopy improves adenoma detection, but real-world results have been inconsistent — until now. A large pragmatic quality improvement trial called CADeNCE, conducted across Veterans Health Administration (VA) facilities, found that making AI-based computer-aided detection (CADe) devices available during colonoscopy significantly boosted adenoma detection rates (ADRs) compared to facilities without the technology.
The study analyzed over 334,000 colonoscopies performed by 816 endoscopists across 139 VA facilities. CADe sites saw ADR climb from 50.7% before device deployment to 54.9% after, while control sites saw a slight dip from 51.8% to 51.1%. Crucially, the benefit was seen across all endoscopist performance levels — even those who were already high performers improved with CADe access.
By the Numbers
Why it matters: Based on these findings, the VA has since rolled out CADe devices to all its facilities. While the impact on long-term cancer outcomes and concerns about endoscopist deskilling remain open questions, this study makes a strong real-world case for broader CADe adoption in colonoscopy programs.
