LAUSR

BACKGROUND AND AIMS Recent meta-analysis showed that up to 26% of adenoma could be missed during colonoscopy. We investigated whether the use of artificial intelligence (AI) assisted real-time detection could provide new insights into mechanisms underlying missed lesions during colonoscopy. METHODS A validated real-time deep learning AI model for detection of colonic polyps was first tested in the videos of tandem colonoscopy of the proximal colon for missed lesions. The real-time AI model was then prospectively validated in total colonoscopy in which endoscopist was blinded to the real-time AI findings. Segmental unblinding of the AI findings were provided and that colonic segment would be re-examined when there were missed lesions detected by AI but not the endoscopist. All polyps were removed for histological examination as the criterion standard. RESULTS Sixty-five videos of tandem examination of the proximal colon were reviewed by AI. AI could detect 79.1% (19/24) of missed proximal adenoma in the video of the first-pass examination. In the 52 prospective colonoscopies, real-time AI detection could detect at least one missed adenoma in 14 (26.9%) patients and increased total number of adenomas detected by 23.6%. Multivariable analysis showed that missed adenoma(s) was more likely when there were multiple polyps (adjusted OR, 1.05; 95% CI, 1.02-1.09; p < 0.0001) or colonoscopy by less experienced endoscopists (adjusted OR, 1.30; 95% CI, 1.05-1.62; p=0.02). CONCLUSION Our findings provide new insights on the prominent role of human factors, including inexperience and distraction, play on missed colonic lesions. With the use of real-time AI assistance, up to 80% of missed adenoma could be prevented.