Validation of artificial intelligence-based bowel preparation assessment in screening colonoscopy (with video)

Liwen Yao,Huizhen Xiong,Qiucheng Li,Wen Wang,Zhifeng Wu,Xia Tan,Chaijie Luo,Hang You,Chenxia Zhang,Lihui Zhang,Zihua Lu,Honggang Yu,Honglei Chen
DOI: https://doi.org/10.1016/j.gie.2024.04.015
Abstract:Background and aims: Accurate bowel preparation assessment is essential for determining colonoscopy screening intervals. Patients with suboptimal bowel preparation are at a high risk of missing >5 mm adenomas and should undergo an early repeat colonoscopy. In this study, we used artificial intelligence (AI) to evaluate bowel preparation and validated the ability of the system to accurately identify patients who are at high risk of having >5 mm adenomas missed due to inadequate bowel preparation. Methods: This prospective, single-center, observational study was conducted at the Eighth Affiliated Hospital, Sun Yat-sen University, from October 8, 2021, to November 9, 2022. Eligible patients who underwent screening colonoscopy were consecutively enrolled. The AI assessed bowel preparation using the e-Boston Bowel Preparation Scale (e-BBPS) while endoscopists made evaluations using BBPS. If both BBPS and e-BBPS deemed preparation adequate, the patient immediately underwent a second colonoscopy; otherwise, the patient underwent bowel re-cleansing before the second colonoscopy. Results: Among the 393 patients, 72 adenomas >5 mm in size were detected; 27 adenomas >5 mm in size were missed. In unqualified-AI patients, the >5 mm adenoma miss rate (AMR) was significantly higher than in qualified-AI patients (35.71% vs 13.19% [P = .0056]; odds ratio [OR], .2734 [95% CI, .1139-.6565]), as were the AMR (50.89% vs 20.79% [P < .001]; OR, .2532 [95% CI, .1583-.4052]) and >5 mm polyp miss rate (35.82% vs 19.48% [P = .0152]; OR, .4335 [95% CI, .2288-.8213]). Conclusions: This study confirmed that patients classified as inadequate by AI exhibited an unacceptable >5 mm AMR, providing key evidence for implementing AI in guiding bowel re-cleansing and potentially standardizing future colonoscopy screening. (Clinical trial registration number: NCT05145712.).
What problem does this paper attempt to address?