Artificial intelligence in pediatric endoscopy for hereditary polyposis syndromes: promises and challenges

Children with hereditary polyposis syndromes require long-term endoscopic surveillance to reduce risks of gastrointestinal complications: malignancy, bleeding, and obstruction, among others. However, surveillance remains challenging because of variable lesion distribution, subtle morphology, interobserver variability, and the burden of repeated procedures throughout the patient's lifetime. Artificial intelligence has transformed adult endoscopy, but its application in pediatric polyposis is less defined. This review summarizes current evidence and explores potential applications for children with hereditary polyposis syndromes.

In adult colonoscopy, meta-analyses and randomized trials demonstrate that artificial intelligence improves adenoma detection rates (ADR) and reduces missed lesions, though evidence linking these benefits to meaningful reductions in colorectal cancer incidence remains insufficient. In capsule endoscopy, artificial intelligence demonstrates high sensitivity for lesion detection while substantially reducing review time. Pediatric data remain limited, with no studies specific to hereditary polyposis syndromes, but early studies support feasibility. Potential applications include automated polyp detection, localization, burden quantification, and longitudinal comparisons across surveillance examinations.

Artificial intelligence has significant potential to improve polyp detection, diagnostic consistency, accuracy, efficiency, and longitudinal disease monitoring in pediatric hereditary polyposis syndromes. Development of pediatric-specific datasets and prospective, multicenter, and outcome-driven validation studies will be essential before widespread clinical implementation.