Clinical and cost-effectiveness of artificial intelligence-guided bowel preparation in improving bowel preparation adequacy among older adults undergoing colonoscopy. An international multi-centre randomised controlled trial.

Objectives To evaluate the clinical efficacy and cost-effectiveness of an artificial intelligence (AI)-guided bowel preparation system in improving bowel preparation adequacy among older adults undergoing colonoscopy. Hypothesis to be tested Participants in the AI group will have higher bowel preparation adequacy rate and a more favourable incremental cost-effectiveness ratio (ICER) compared with standard practice (SP). Design and subjects Older adults (≥65 years) undergoing colonoscopy who require a standard high-volume polyethylene glycol (PEG) regimen and have access to smartphones will be recruited from multiple centres in Hong Kong, China, Singapore, and Australia. Participants will be randomized in a 1:1 ratio to either the AI or SP groups. Study instruments The SP group will receive the standard high-volume PEG regimen and complete the full prescribed dose before colonoscopy. In the AI group, in addition to SP, an AI system will be used to predict bowel preparation adequacy by analysing rectal effluent images. Participants will discontinue PEG consumption once the AI system yields a 'pass' result or after completing the full dose. Outcome measures The primary outcome is bowel preparation adequacy. Secondary outcomes include the incremental ICER and other colonoscopy quality indicators in both groups. Data analysis Appropriate statistical tests will be used to evaluate outcomes. ICER will be calculated by dividing the mean total cost difference between groups by the difference in the proportion of early repeat colonoscopies within one year. Expected results The AI-guided system is expected to improve bowel preparation adequacy and be more cost-effective than SP.