Oculomics in Gastrointestinal Disorders: A narrative review

We reviewed various ocular biomarkers relevant to gastrointestinal (GI) disorders and examined the potential role of artificial intelligence (AI) in detecting, integrating, and translating these biomarkers into clinical gastroenterology practice.

A literature search from PubMed, MEDLINE, Scopus, Google Scholar, and Web of Science from the year 2000 up to February 2026 was conducted. We included published articles on ocular findings in gastrointestinal disorders and/or AI-based analysis of ocular biomarkers with gastrointestinal connections. Results suggested a positive correlation between AI-based ocular biomarkers and GI disorders.

After reviewing the literature, we found that the most frequently described ocular biomarkers were retinal vascular caliber and arteriovenous ratio on fundus photography; retinal nerve fibre layer and choroidal changes on optical coherence tomography; and vessel-density changes on optical coherence tomography angiography. The persistent signal came from inflammatory bowel disease and steatotic liver disease. AI may improve image processing, feature extraction, and pattern recognition, but more multicentric studies are needed to establish definitive analogy.

This review suggests that ocular biomarkers could provide a simple, non-invasive way to identify and monitor gastrointestinal diseases, especially inflammatory bowel disease and steatotic liver disease. Changes in retinal blood vessels, the retinal nerve fibre layer, the choroid, and OCTA-derived vessel density may reflect the inflammation, metabolic disturbances, and small-vessel changes associated with these conditions. However, these findings should be interpreted with caution because many existing studies included small or selected patient groups, used cross-sectional designs, applied different imaging methods, and did not consistently account for factors such as age, vascular risk, medications, or coexisting eye disease. Artificial intelligence may improve the analysis of complex eye images by identifying subtle patterns. Still, high model performance alone does not prove that these tools are accurate, reliable, or specific enough for clinical use. Future studies should therefore include large, multicentre, prospective populations, standardized imaging and reporting methods, independent external validation, and clinically meaningful outcomes. These steps will help determine whether AI-assisted oculomics can reliably support gastrointestinal screening, disease classification, risk assessment, and long-term monitoring.

Various ocular biomarkers offer a promising, non-invasive approach to studying the deeper nuances of gastrointestinal disease. AI may help scale early biomarker detection and avoid gaps in analysis. Still, routine clinical use will require better external validation, clearer reporting standards, and prospective translational studies.