Research advances in predictive models for inflammatory bowel disease: a review

Inflammatory Bowel Disease (IBD) is a group of chronic intestinal inflammatory disorders characterized by complex disease progression and significant individual heterogeneity, posing significant challenges for clinical management. Conventional assessment methods exhibit limitations in early diagnosis, objective quantification of disease activity, and prediction of treatment response. Emerging predictive models integrating clinical biochemistry, genomics, proteomics, metabolomics, and advanced radiomics have offered novel paradigms for personalized medicine in IBD, with the rapid evolution of artificial intelligence (AI) serving as a critical technological pillar for algorithmic orchestration. Nevertheless, widespread implementation of current IBD models in real-world clinical settings is substantially hindered by their dependency on cross-sectional snapshots, insufficient phenotypic stratification of ulcerative colitis and Crohn’s disease, and a lack of robust prospective causal verification. This article aims to review the latest research progress of both traditional and AI-driven predictive models in the field of IBD, analyzing their practical value in diagnostic classification, activity scoring, individualized response forecasting, and complication warning systems. The goal is to provide a foundation for breaking current clinical bottlenecks, establishing a “prediction-intervention-reevaluation” clinical framework, and facilitating the translation of predictive models into practical clinical tools.