S657 Artificial Intelligence-Assisted Colonoscopy Improves Adenoma Detection: Meta-Analysis of 15 Randomized Trials in Colorectal Screening

Introduction: Adenoma detection rate (ADR) is a cornerstone quality indicator for colorectal cancer prevention. Despite rigorous technique, traditional colonoscopy may miss up to 26% of adenomas. Artificial intelligence–assisted colonoscopy (AI-C) offers real-time image augmentation and may improve detection metrics. We conducted a meta-analysis of randomized controlled trials (RCTs) comparing AI-C to conventional colonoscopy (C-C) for ADR, polyp detection rate (PDR), missed adenoma rate (MAR), and other performance indicators. Methods: This PRISMA-guided meta-analysis included 15 RCTs (n = 11,038 patients) from 2000–2025. Searches spanned PubMed, Cochrane, Embase, and Google Scholar. Eligible trials compared AI-C with C-C during screening or surveillance colonoscopy in adults. Pooled risk ratios (RR) or mean differences (MD) were calculated using random-effects models. Outcomes included ADR, PDR, MAR, and withdrawal time. Heterogeneity was assessed via I² statistics. Risk of bias was evaluated using Cochrane ROB-2. Results: ADR: Significantly higher with AI-C (RR 1.27; 95% CI: 1.14–1.41; I² = 36%). PDR: Improved across 12 studies (RR 1.21; 95% CI: 1.09–1.35; I² = 33%). MAR: Lower in AI arms (RR 0.58; 95% CI: 0.45–0.74). Withdrawal Time: Marginal increase (+0.41 minutes; 95% CI: –0.02 to +0.84; I² = 42%). Adverse Events: Comparable across groups, with no significant increase in perforation or post-polypectomy bleeding. Conclusion: AI-assisted colonoscopy significantly enhances adenoma and polyp detection while preserving procedural safety. These findings support integrating AI into routine colonoscopy practice to standardize quality and minimize miss rates, particularly in high-risk populations.