Artificial intelligence in Latin American hepatology: current evidence, implementation barriers, and strategic priorities.

Authors

Ezequiel Ridruejo, Andreas Teufel, Mario Alvares-da-Silva, Juan Turnes

Published in

Annals of hepatology. Pages 102432. Aug 30, 2026. Epub Aug 30, 2026.

Abstract

Chronic liver diseases are an increasing cause of morbidity and mortality in Latin America, driven by the convergence of alcohol and metabolic-associated steatotic liver disease, and viral hepatitis. The region faces structural barriers including fragmented data systems, delayed diagnosis, disparities in access, and limited research capacity. Although artificial intelligence has shown clinical utility across hepatology, evidence supporting its implementation in Latin America remains limited. We reviewed the AI applications most relevant to regional hepatology, the barriers to adoption, and priorities for safe implementation. We conducted a narrative review based on iterative appraisal of peer-reviewed studies, regional epidemiological reports, and digital health policy documents identified through March 2026. Priority was given to evidence reporting external validation, pragmatic clinical testing, or direct relevance to implementation in regional health systems. Discriminative AI is most mature in fibrosis risk stratification, digital pathology for MASH trials, and opportunistic case finding using electrocardiogram-based screening. In a pragmatic cluster-randomized trial of 15,596 adults, AI-enabled electrocardiography doubled the detection of previously undiagnosed advanced chronic liver disease in primary care. Retrieval-augmented generation improves the accuracy of large language models for hepatitis C guideline interpretation, but generative systems still pose hallucination risk. Adoption in Latin America is limited by uneven digital maturity, weak interoperability, scarce local validation, regulatory fragmentation, and insufficient workforce preparation. AI may help address major challenges in Latin American hepatology, but implementation should be phased: first infrastructure and governance, then locally validated discriminative tools, and finally supervised generative AI.

PMID:
42669326
Bibliographic data and abstract were imported from PubMed on 31 Aug 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.