Acute pancreatitis (AP) is one of the most common diseases of the gastrointestinal tract, which in 20% of cases can turn into a severe form, with mortality reaching up to 30%. One of the cornerstones of AP treatment is early nutritional treatment. Feeding intolerance (FI) occurs in up to 25% of patients with AP and is associated with a more severe disease course and poorer clinical outcome. Feeding intolerance can have a multifaceted clinical presentation. The early identification of FI risk factors and appropriately conducted nutritional treatment are critical to the course of the disease. In this review, we summarize the current knowledge of feeding intolerance in AP, its pathomechanisms and risk factors, and its impact on disease progression. We also present suggestions for the management of feeding intolerance.

Acute pancreatitis (AP) is one of the most common gastrointestinal diseases and is characterized by a highly variable clinical course ranging from mild self-limiting inflammation to severe disease associated with persistent organ failure and high mortality. Early risk stratification is essential for timely therapeutic decision-making and improved patient outcomes. Although conventional prognostic scoring systems remain widely used, their predictive performance is limited by moderate accuracy, delayed applicability and the requirement for numerous clinical or imaging parameters. Recent advances in artificial intelligence (AI) and machine learning (ML) have enabled the development of predictive models capable of integrating demographic, clinical, laboratory and imaging data to improve early outcome prediction. This review summarizes current evidence on the application of AI and ML in AP, focusing on the prediction of disease severity, organ failure, intensive care unit admission and mortality. We also discuss explainable AI approaches and the methodological strengths and limitations of existing models, including validation, calibration, overfitting and reproducibility. Finally, we highlight future perspectives, including multimodal AI, prospective clinical validation and regulatory considerations that should facilitate the safe integration of AI-based decision-support tools into routine clinical practice.