Artificial intelligence in multidisciplinary and cross-sector public health practice: A scoping review

This scoping review examined how artificial intelligence is applied in multidisciplinary and cross-sector public health practice, focusing on public health roles, collaborative implementation, and cross-sector governance and equity risks.

Scoping review.

Following Arksey and O'Malley's framework and PRISMA-ScR reporting guidance, six databases were searched. Sources were included if they examined artificial intelligence or artificial intelligence-enabled decision support in public, population, or community health contexts and provided information on public health functions, collaboration, implementation, equity, or governance. Data were charted and synthesised narratively by study characteristics, artificial intelligence roles, collaborators, implementation conditions, and governance risks.

27 sources published between 2021 and 2026 were included. Most were reviews, evidence syntheses, policy or guidance documents, or conceptual papers, with relatively few empirical or practitioner-oriented studies. Artificial intelligence was mainly used for surveillance and early warning, population risk modelling, health communication, misinformation detection, policy intelligence, participatory co-creation, and routine public health operations. Implementation depended on collaboration among public health practitioners, epidemiologists, data scientists, policymakers, legal and ethics experts, communication specialists, community organisations, and affected populations. Cross-sector governance and equity risks included poor data quality, limited interoperability, workforce gaps, weak real-world evaluation, privacy risks, and unclear accountability.

Artificial intelligence applications in public health are expanding, but the evidence base remains heterogeneous and empirically limited. Routine operational support and AI readiness, together with surveillance and early warning, were among the more frequently represented roles, while empirical evidence on real-world implementation, community engagement and governance evaluation remained limited. Future research should move beyond model performance to examine workflow fit, equity impact, public trust, accountability and population-level outcomes.