Aims
Artificial intelligence (AI) is increasingly used in GI endoscopy to enhance polyp and adenoma detection. We measured the AI enhanced endoscopy on adenoma miss rate (AMR), polyp miss rate (PMR) and sessile serrated lesion miss rate (SSL‑MR).
Methods
We searched PubMed and Embase databases to July 2025 for RCTs comparing AI‑assisted with standard Endoscopy. Pooled risk ratios (RR) were calculated with random effects.
Results
Twelve trials showed significantly AI endoscopy lowered AMR by 39 % (RR 0.61, 95 % CI 0.5‑0.75) and PMR by 36 % (RR 0.64, 95 % CI 0.44‑0.91). SSL‑MR fell but was not significant (RR 0.58, 95 % CI 0.29‑1.16).
Conclusions
Our meta-analysis showed that AI-assisted endoscopy results in a significant reduction in missed adenomas and polyps. A similar reduction was also observed for missed sessile serrated lesions but did not reach significance. These results point to the potential of AI enhancement to improve the accuracy of endoscopic surveillance by catching more lesions on the first pass. Widespread use of AI in endoscopy could make fewer missed lesions and reduce the incidence of interval neoplasms.