S658 Artificial Intelligence-Enhanced Colonoscopy Increases Adenoma and Small Polyp Detection: A Meta-Analysis of 13 RCTs

Introduction: Colorectal cancer (CRC) is the second most commonly diagnosed malignancy and ranks third in cancer-related mortality in the United States. Early detection through colonoscopy is a cornerstone of CRC prevention. The adenoma detection rate (ADR) is a validated quality metric, with higher ADRs linked to decreased incidence of post-colonoscopy CRC. Artificial intelligence (AI) has emerged as a promising adjunct in endoscopy, capable of enhancing detection by reducing human error and highlighting subtle or overlooked lesions. While several randomized controlled trials (RCTs) have demonstrated the utility of AI, a comprehensive meta-analytic comparison between AI-assisted and standard colonoscopy (SC), especially in detecting small adenomas (≤5 mm), remains lacking. Methods: We conducted a systematic review and meta-analysis of RCTs published between 2019 and 2024. Databases searched included PubMed, Google Scholar, ScienceDirect, Cochrane, and PLOS One. Thirteen RCTs involving 11,521 patients were included (5,585 AI-assisted; 5,636 SC). Data were pooled using odds ratios (OR) with 95% confidence intervals (CIs), and heterogeneity was assessed with the I² statistic. Subgroup analysis focused on the detection of diminutive lesions (≤5 mm). Results: AI-assisted colonoscopy demonstrated a significantly higher overall ADR than SC (OR 1.43; 95% CI: 1.28–1.60; I² = 43%; P < 0.00001). In detecting small adenomas and polyps (≤5 mm), AI also showed superior performance (OR 1.56; 95% CI: 1.24–1.97; I² = 73%; P = 0.0002). These results reflect a robust and consistent benefit of AI integration across key outcome measures. Conclusion: AI-enhanced colonoscopy significantly improves the detection of adenomas, including diminutive lesions, compared to standard methods. These findings support the integration of AI into routine CRC screening protocols and justify continued research into AI-based diagnostic tools to advance precision endoscopy and reduce cancer burden.