# Artificial Intelligence in Pediatric Inflammatory Bowel Disease: Applications in Diagnosis, Monitoring, and Therapeutic Decision-Making

## TL;DR

Artificial intelligence shows promise in improving diagnosis and treatment of pediatric inflammatory bowel disease, but more research is needed for pediatric-specific validation.

## Contribution

The paper provides a critical review of AI applications in pediatric IBD, highlighting the need for pediatric-specific validation and integration into clinical care.

## Key findings

- AI models using machine learning and deep learning show promising performance in analyzing endoscopic, histological, and imaging data for pediatric IBD.
- Multimodal AI approaches integrating clinical and biomarker data outperform unimodal models in disease monitoring and therapeutic response prediction.
- Emerging AI applications in digital biomarkers and telemedicine enable continuous, non-invasive disease assessment in pediatric IBD.

## Abstract

What are the main findings?
Artificial intelligence applications in pediatric inflammatory bowel disease show promising performance, particularly in image-based and multimodal assessment.Pediatric-specific evidence remains limited, with many studies relying on adult-derived data.

Artificial intelligence applications in pediatric inflammatory bowel disease show promising performance, particularly in image-based and multimodal assessment.

Pediatric-specific evidence remains limited, with many studies relying on adult-derived data.

What are the implications of the main findings?
Artificial intelligence may support standardized, non-invasive, and longitudinal disease assessment in pediatric inflammatory bowel disease.Clinical translation requires pediatric validation, explainable models, and integration into care pathways.

Artificial intelligence may support standardized, non-invasive, and longitudinal disease assessment in pediatric inflammatory bowel disease.

Clinical translation requires pediatric validation, explainable models, and integration into care pathways.

Background: Pediatric inflammatory bowel disease (IBD) is characterized by a heterogeneous and often aggressive disease course, requiring complex multimodal assessment and long-term monitoring. Artificial intelligence (AI) has emerged as a promising tool to support clinical decision-making by enabling an objective analysis of large, multidimensional datasets. Objectives: This narrative review aims to critically synthesize current evidence on the application of AI across the diagnosis, monitoring, and treatment of pediatric IBD. Methods: A narrative literature review was conducted using the PubMed (MEDLINE) and Cochrane Library databases, including publications available up to December 2025. Pediatric-focused studies were prioritized. However, due to the limited availability of pediatric-specific AI research, a considerable proportion of the evidence reviewed derives from adult or mixed cohorts, which were included when methodological frameworks or clinically relevant endpoints were applicable to pediatric IBD. Eligible publications included narrative and systematic reviews, observational studies, and clinical trials focusing on AI applications in endoscopy, histology, imaging, disease monitoring, and therapeutic response prediction. Results: AI-based models, particularly those using machine learning and deep learning, demonstrated promising performance in the automated analysis of endoscopic, histological, and imaging data, reducing interobserver variability and improving workflow efficiency. Multimodal approaches integrating imaging, clinical, and biomarker data consistently outperformed unimodal models. Emerging applications in patient-centered monitoring, digital biomarkers, and telemedicine enabled continuous disease assessment and early detection of flares, with particular relevance in pediatric settings where repeated, non-invasive monitoring is essential. AI-driven models also showed promising accuracy in predicting therapeutic response, supporting treatment stratification and precision medicine strategies. Conclusions: AI shows promising potential to complement clinical expertise in pediatric IBD by supporting diagnostic assessment, disease monitoring, and therapeutic optimization. However, translation into routine clinical practice remains constrained by methodological heterogeneity, limited pediatric-specific validation, and unresolved ethical and regulatory challenges. Future research should prioritize prospective multicenter pediatric studies, the development of transparent and explainable models, and the integration of AI-based tools into clinically meaningful and patient-centered care pathways.