JMIR Publications

ABSTRACT

The 2023 Global Burden of Disease study highlights substantial differences in life expectancy across Asia, with a 17-year gap between the highest- and lowest-performing countries. These disparities reflect varying trends in cardiovascular disease, metabolic risks, and maternal and child health. Although the data are available, policymakers often lack the capacity to translate them into timely, detailed action. This analytical gap is where agentic artificial intelligence (AI) could make a meaningful contribution. Unlike conventional static models, AI agents can independently plan, retrieve, synthesise, and communicate information across multiple tasks. Conversational interfaces for querying complex health databases in local languages, automated evidence synthesis, causal reasoning to explore potential mechanisms, and biosurveillance systems are among the capabilities that integrate diverse real-world signals in near real-time. Collectively, these tools could enable rural health authorities to analyse local mortality data with capabilities previously limited to well-resourced institutions. However, these advances also carry significant risks, including fabricated estimates, automation bias in clinical and policy decisions, language-related inequalities, and data sovereignty concerns arising from external processing of local health information without adequate human oversight. Therefore, the global health community should establish minimum requirements before large-scale deployment including transparent benchmarking, comprehensive auditability, mandatory human oversight, investment in local capacity, robust foundational data infrastructure, and inclusive multilateral governance. Whether agentic AI promotes health equity or amplifies errors ultimately depends on how rigorously these safeguards are implemented.