Artificial intelligence in non-invasive colorectal cancer screening: current status, controversies, and future directions
Colorectal cancer (CRC) is a leading cause of cancer-related death worldwide, and early detection remains the most effective strategy for reducing its incidence and mortality. Traditional screening modalities are limited by suboptimal adherence, invasiveness, and inequitable access. The rapid evolution of artificial intelligence (AI) has opened new avenues for non-invasive CRC screening, encompassing blood-based liquid biopsy, multi-omic stool analysis, breath volatilomics, tongue-image analytics, and radiomic interpretation of medical imaging. A growing number of AI-driven models have shown promising diagnostic performance for CRC and its precursor lesions; selected studies have reported area-under-the-curve values above 0.95, although performance varies substantially according to modality, population, lesion stage, and validation design. Nevertheless, the clinical value of AI-assisted screening remains the subject of considerable debate. Discordant findings on adenoma detection rates, increased detection of non-neoplastic polyps, performance variability across anatomical sites and population subgroups, and concerns regarding model generalizability and equity continue to challenge widespread translation. This mini-review summarizes recent advances in AI-enabled non-invasive CRC screening, critically examines current controversies, identifies key research gaps, and discusses promising future directions, including multi-modal data integration, explainable AI, lifecycle disease management, and equitable global deployment.
Authors
- Jiang Liu (ORCID: https://orcid.org/0000-0002-0654-707X)
- Minxia Shan
Institutions
- Shaoxing People's Hospital (CN)
Publication Details
- Journal
- Frontiers in Oncology
- Published
- 2026-11-20
- DOI
- https://doi.org/10.3389/fonc.2026.2004675
- Primary Topic
- Colorectal Cancer Screening and Detection
- Type
- article
- Field-Weighted Citation Impact
- 0.00