A critical perspective and integrative PRISM framework on artificial intelligence and the future of public health

Abstract

Artificial intelligence is transforming public health at an unprecedented pace – accelerating disease surveillance, reshaping evidence synthesis, augmenting clinical and policy decision-making, and fundamentally altering how populations interact with health information. Yet this rapid integration has consistently outpaced the development of ethical governance, equitable implementation, and rigorous evaluative frameworks capable of ensuring that AI serves – rather than undermines – public health’s foundational commitment to health equity. This article presents a critical, forward-looking thematic synthesis of the published evidence on AI in public health, examining conceptual foundations and definitional clarity; current applications across surveillance, evidence synthesis, diagnostics, health communication, and resource allocation; ethical, legal, and socio-political dimensions including algorithmic bias, data privacy, transparency, accountability, sycophancy, cognitive homogenisation, commercialisation, and environmental sustainability; equity, diversity, and global representation with particular attention to the digital divide, WEIRD bias, and Indigenous data sovereignty; and implementation challenges spanning workforce readiness, data governance, regulatory fragmentation, public trust, and methodological limitations. Drawing on this synthesis, the article proposes the PRISM Framework – a comprehensive, five-dimensional, evidence-grounded model comprising Participatory Design and Community Engagement, Responsible and Regulatory Governance Infrastructure, Inclusive and Equitable Representation, Scientific Rigour and Scalable Infrastructure, and Monitoring, Evaluation, and Accountability. PRISM provides an integrative architecture for equitable, accountable, and evidence-based AI adoption in public health, addressing the critical gap between existing partial frameworks and the multidimensional reality of responsible AI deployment at population scale. The article concludes with structured recommendations for researchers, public health professionals, and policymakers, alongside a future research agenda oriented toward ensuring that AI genuinely advances health equity across diverse populations globally.