Endo-Hepatology and Artificial Intelligence: New Avenues for Diagnosis and Intervention

Artificial intelligence (AI) is gaining momentum in the field of endo-hepatology, offering potential improvements in diagnosis, risk stratification, and procedural outcome prediction. This review outlines AI applications across endoscopic domains including variceal screening, endoscopic retrograde cholangiopancreatography (ERCP), cholangioscopy, and metabolic interventions, with particular promise seen in image-based models and radiomics. While early studies-mostly retrospective and pilot in nature-demonstrate improved accuracy and efficiency over conventional methods, only a few have progressed to prospective validation or real-world clinical integration. Notably, AI has shown the capability to reduce unnecessary endoscopies for variceal screening, predicting post-ERCP complications, and augmenting endoscopic ultrasound (EUS) training. However, the integration of AI into EUS-guided interventions, which are the core interventions in endo-hepatology, remains underdeveloped due to challenges like inconsistent data and difficulties with real-time image interpretation. While the outlook is encouraging, the current evidence calls for cautious optimism. Broad clinical adoption will depend on further prospective studies, validation across diverse settings, and stronger collaboration between clinicians and data scientists to ensure AI tools are reliable, interpretable, and clinically meaningful.