7th Global Gastroenterology and Artificial Intelligence Summit (Live)

10 - Dominitz

Pdf Summary

The presentation, led by Dr. Jason A. Dominitz and the Veterans Health Administration's National Gastroenterology and Hepatology Program (NGHP), discusses the large-scale implementation and evaluation of Artificial Intelligence (AI)-based Computer-Aided Detection (CADe) technology in colonoscopy services across the Veterans Affairs (VA) healthcare system.<br /><br />While randomized controlled trials (RCTs) had demonstrated CADe's benefit in improving adenoma detection rates (ADR), observational studies were inconsistent. To address these uncertainties, the NGHP secured funding and established a national VA contract in 2022 to acquire CADe devices (specifically the GI Genius system) and strategically distributed these to select VA facilities prioritizing those with lower baseline ADR and wide geographic representation.<br /><br />The CADeNCE (Computer-Aided Detection of Neoplasia During Colonoscopy Evaluation) program evaluated CADe’s impact by comparing colonoscopy outcomes pre- and post-deployment across CADe-equipped and control sites. Surveys of 124 endoscopists using CADe showed that most found procedures neither significantly longer nor shorter, and many reported that CADe modestly increased their ADR and sessile serrated lesion detection. However, concerns included false positives, distraction, potential overreliance on technology, and cost.<br /><br />Objective data from over 75,000 post-deployment colonoscopies revealed statistically significant increases in ADR at CADe sites (from 50.7% to 54.9%) compared to control sites, without prolonging withdrawal time or changing adenocarcinoma detection. Multivariable models confirmed a 21% higher odds of adenoma detection with CADe. Benefits were seen across providers regardless of their baseline ADR.<br /><br />Following successful initial waves, 385 CADe units were distributed to 153 VA facilities. The program integrated CADe use tracking into the Provation MD system used for documenting GI procedures across the VA, enabling ongoing centralized data collection to monitor usage patterns and outcomes, including post-colonoscopy colorectal cancer rates.<br /><br />In summary, the VA’s systematic rollout of AI CADe in colonoscopy showed feasibility, provider acceptance, and measurable improvements in adenoma detection, supporting broader adoption while continuing rigorous outcome monitoring.

Keywords

Artificial Intelligence

Computer-Aided Detection

CADe

Colonoscopy

Adenoma Detection Rate

Veterans Affairs Healthcare System

GI Genius System

CADeNCE Program

Endoscopist Surveys

Post-Colonoscopy Outcomes