Artificial Intelligence-Assisted Colonoscopy for Colorectal Cancer Screening: A Multicenter Randomized Controlled...

Area of Science:

• Gastroenterology
• Medical Technology
• Clinical Trials

Background:

• Large-scale multicenter randomized controlled trials (RCTs) are lacking for AI-assisted colonoscopy in asymptomatic populations.
• AI-assisted colonoscopy has shown promise in improving polyp detection and characterization.

Purpose of the Study:

• To compare the efficacy of AI-assisted colonoscopy versus conventional colonoscopy for adenoma detection in asymptomatic individuals undergoing colorectal cancer screening.
• To evaluate secondary outcomes including the number of adenomas detected and the impact of endoscopist experience.

Main Methods:

• A multicenter RCT involving 3059 asymptomatic subjects aged 45-75 years across 6 referral centers.
• Participants were randomized to either AI-assisted colonoscopy (using the Eagle-Eye system) or conventional colonoscopy.
• The primary outcome was the overall adenoma detection rate (ADR).

Main Results:

• AI-assisted colonoscopy demonstrated a significantly higher overall ADR (39.9% vs 32.4%, P < .001) compared to conventional colonoscopy.
• Advanced ADR and ADR for both expert and nonexpert endoscopists were also significantly improved with AI assistance.
• The mean number of adenomas per colonoscopy was higher in the AI group, with a slightly longer withdrawal time.

Conclusions:

• AI-assisted colonoscopy is effective in improving adenoma detection rates, including advanced adenomas, in asymptomatic patients.
• The benefits were observed across both expert and nonexpert endoscopists, suggesting broad applicability.
• This study provides robust evidence from a large multicenter RCT supporting the use of AI in colonoscopy for cancer screening.