S308 No Improvement in Endoscopist Performance With Computer-Aided Colonoscopy in Real-World Community Settings

Introduction: Computer-aided detection (CADe) systems, such as GI Genius, have shown promise in improving adenoma detection rate (ADR) in controlled settings. However, the effectiveness of CADe in real-world, community-based colonoscopy practice remains unclear. This study aimed to evaluate whether CADe improves ADR and adenoma per colonoscopy (APC) in a community endoscopy center. Methods: This was a retrospective cohort study at a single community medical center. Colonoscopies performed from April 2022 to April 2024 were reviewed using the GIQuIC database. GI Genius was implemented in mid-April 2023. Colonoscopies performed 1 year before installation (April 2022–April 2023; non-CADe group) were compared with those performed 1 year after installation (mid-April 2023–April 2024; CADe group). The primary outcome was ADR; secondary outcomes included APC and serrated lesion detection rate. Statistical analysis was performed using JMP v10. Results: A total of 10,125 colonoscopies were included (non-CADe: 4,891; CADe: 5,233). Baseline demographics, bowel preparation quality (BBPS >2), and withdrawal times were similar between groups. Median insertion time was longer in the CADe group. ADR did not significantly differ: 55% in the non-CADe group vs 56% in the CADe group (RR, 0.98; 95% CI, 0.88–1.22). Serrated adenoma detection was also similar (13% vs 14%, RR, 1.06; 95% CI, 0.85–1.46). However, advanced adenoma detection was higher in the CADe group. Conclusion: In this real-world community setting, CADe-assisted colonoscopy did not significantly improve ADR or serrated lesion detection. The clinical utility of CADe systems outside academic centers remains uncertain and warrants further prospective investigation.