AI-enabled multi-omics integration in colorectal cancer: from molecular stratification to clinical translation

TLDR

It is argued that, despite encouraging advances in biomarker discovery and risk prediction, most current studies remain retrospective and are constrained by heterogeneity of data sources, limited standardisation, weak interpretability, and insufficient external or prospective validation.

Abstract

Colorectal cancer (CRC) remains a heterogeneous disease for which improved molecular stratification is needed across the clinical pathway. Multi-omics technologies have expanded insight into CRC biology, and artificial intelligence (AI) has created new possibilities for integrating molecular, pathological, imaging, and clinical data. This review examines how these approaches are being applied across screening, diagnosis, treatment, and prognosis, with particular emphasis on their clinical relevance and translational limitations. We argue that, despite encouraging advances in biomarker discovery and risk prediction, most current studies remain retrospective and are constrained by heterogeneity of data sources, limited standardisation, weak interpretability, and insufficient external or prospective validation. AI-enabled multi-omics integration has substantial potential in CRC, but meaningful clinical impact will require rigorous validation and implementation frameworks suited to routine care.