Artificial intelligence (AI) is becoming increasingly significant in healthcare systems because of its efficiency in handling complicated datasets, identifying disease trends, and facilitating faster and standardized decision-making. In complex conditions such as inflammatory bowel disease (IBD), where care depends on integrating clinical, endoscopic, histological, imaging, and multi-omics data, AI offers a way to bring these diverse inputs together and guide more personalized, timely, and outcome-focused treatment decisions.
IBDs, including Crohn's disease (CD) and ulcerative colitis (UC), are complex, long-term conditions, with symptoms widely varying among patients. For clinicians, assessing disease activity and planning treatment often requires integrating clinical, endoscopic, histological, imaging, and multi-omics data. This makes IBD a strong candidate for AI-driven support, especially when the goal is to move from generalized care toward precision medicine.
In a new review article, researchers from University College Cork, Ireland, led by Professor Marietta Iacucci, summarize the growing evidence on how AI is being applied across the full spectrum of IBD care. It reviews developments in endoscopy, histology, cross-sectional imaging, digital health, multi-omics, and clinical decision support, while also outlining the barriers that still stand between promising models and routine clinical use. The study was published in the Chinese Medical Journal on June 09, 2026. "IBD is a compelling use case for AI, given its biological and clinical complexity. Indeed, objective disease assessment is often challenging," says Prof. Iacucci.
Endoscopic assessment is one of the central tools used to diagnose IBD and monitor disease activity. Endoscopic scoring can be subjective and affected by interobserver variability, fatigue, and differences in expertise. AI-based systems, particularly deep learning models, may help standardize interpretation by identifying patterns of inflammation, grading disease severity, detecting mucosal healing, and supporting the recognition of dysplasia in patients with long-standing colitis.
For UC, several AI models have been developed to assess endoscopic images or video-based data. Some approaches can evaluate mucosal healing or classify severity using standard scoring systems. In CD, AI has been especially useful in capsule endoscopy and enteroscopy, where models can help detect ulcers, bleeding, strictures, and other small bowel lesions. Such tools may also reduce the time needed to review long capsule endoscopy videos, making workflows more efficient while maintaining accuracy.
The review also highlights the role of AI in histological assessment for IBD. Histological remission is increasingly recognized associated with improved clinical outcomes, but microscopic assessment can be time-consuming and variable. AI-enabled digital pathology may help automate sample evaluation, quantify disease activity, and identify features linked with relapse or treatment response. By combining endoscopic and histological information, future AI models may provide a more complete picture of intestinal healing.
Deep learning can derive quantitative data from cross-sectional imaging techniques like computed tomography enterography, magnetic resonance enterography, and intestinal ultrasound. These tools may characterize inflammation, fibrosis, strictures, and complications, while aiding in the prediction of treatment response. AI-enhanced imaging is crucial for noninvasive monitoring of disease beyond the intestinal surface.
A major theme of the review is the value of multimodal integration. AI can combine clinical data, endoscopy, histology, imaging, biomarkers, microbiome profiles, and other omics data to support patient phenotyping, risk stratification, treatment selection, outcome prediction, and drug discovery. This integrated approach could help clinicians identify which patients are likely to relapse, which may respond to a specific treatment, and which may need closer monitoring.
Beyond hospital-based assessment, the review describes how digital health tools, natural language processing, large language models, wearable devices, home-based biomarkers, and remote monitoring platforms may reshape IBD management. These technologies could help clinicians extract information from medical records, support patient education, monitor symptoms and biomarkers outside the clinic, and identify early warning signs of disease flares. If validated, such tools could make IBD care more proactive and patient-centered.
AI is not yet ready to replace clinical judgment due to limitations such as small datasets, retrospective designs, and lack of external validation. Generalizability is a challenge, and ethical concerns like privacy and bias must be addressed.
The review suggests AI should be a clinical aid rather than an autonomous decision-maker. Future developments should emphasize multicenter validation, standardized datasets, and collaboration among stakeholders to enhance IBD assessment and move toward precision medicine. "Overall, AI has the potential to standardize and elevate IBD care, not as a replacement for clinical judgment, but as a powerful tool for integrating complex, multimodal data, ushering in a new era of personalized, precision medicine-driven IBD management," concludes Prof. Iacucci.
Reference
DOI: http://doi.org/10.1097/CM9.0000000000004170
