ARTIFICIAL INTELLIGENCE ASSISTED COLONOSCOPY IMPROVES PDR AND ADR IN RIGHT COLON IN PATIENTS WITH IBD: A COHORT STUDY IN A HIGH-VOLUME CENTER

Introduction
Artificial Intelligence assisted colonoscopy has been shown to improve adenoma detection rate (ADR) and polyp detection rate (PDR) in screening colonoscopy. The impact of this technology has not been examined in patients with inflammatory bowel disease (IBD). In this study we will examine the impact of AI assisted surveillance colonoscopy in IBD patients.

Methods:
We identified patients with IBD who underwent AI assisted colonoscopy. We used propensity score matching to compare them to a cohort with IBD evaluated by conventional colonoscopy surveillance. Matching was done to account for gender, abdominal and colorectal surgical history (metrics of IBD severity), comorbidities, and family history in this group among AI and non-AI assisted colonoscopies. Primary outcomes were ADR and PDR. Secondary outcomes include ADR in left versus right side of the colon.

Results:
Among 889 patients who underwent AI assisted colonoscopy, 55 patients with IBD were identified. Propensity score matching generated a comparable group (i.e. gender, severity, comorbidities) of 37 IBD patients (2:3 ratio) evaluated by conventional colonoscopy for dysplasia surveillance. We found a significant increase in PDR in IBD patients who underwent AI assisted colonoscopy, (PDR: 54.5% AI vs 27% non-AI, p=0.009). We also found an increase in ADR overall (29% for AI vs 18.9% non-AI) which was statistically significant for right colon neoplasia (32.7% AI vs 13.5%; p=0.037).

Conclusion:
AI assisted colonoscopies may increase PDR and ADR in IBD patients especially on the right side of colon. These findings suggest AI assisted colonoscopy may benefit IBD patients undergoing colon cancer surveillance.