Communications Health Opens Collection on Trustworthy AI in Public Health
Artificial intelligence is moving deeper into public health research and practice, and the central question is no longer only what these tools can do. The sharper question is whether their outputs can be trusted when they influence communication, surveillance, and decision-making across populations.
Communications Health has opened a collection dedicated to that problem, with a submission deadline of 06 August 2027. The collection is framed around trustworthy AI in public health and is tied to SDG 3, the United Nations goal to ensure healthy lives and promote well-being for all at all ages. Its focus is not limited to model accuracy. Reliability, transparency, accountability, fairness, and safety sit at the center of the call.
The framing is deliberate. In public health, an AI system that performs well in a narrow technical test can still fail the people it is meant to serve if its predictions are opaque, biased, unsafe, or poorly integrated into real-world workflows. Surveillance systems, risk stratification tools, and communication platforms all operate in settings where errors can scale quickly. When a tool is deployed across communities, the consequences of poor design are not abstract; they are operational, social, and clinical.
Reliability, Transparency, and Safety Define the Editorial Scope
The collection invites work that examines trustworthiness from several directions at once. One strand concerns AI applications in health communication, disease surveillance, and decision-making. Another asks how trustworthy AI should be developed, validated, and evaluated. A third looks at the human side of adoption: how clinicians, public health teams, and communities respond to these systems, and what shapes whether they are accepted or resisted.
This broad scope reflects a basic reality of public health implementation. A model does not operate in isolation. It sits inside a chain of data collection, interpretation, communication, and action. If any link in that chain is weak, the system may magnify uncertainty rather than reduce it. Fairness also becomes a public health issue rather than a purely technical one, because biased outputs can distort who gets attention, who gets excluded, and which communities are seen as higher or lower priority.
The collection also explicitly welcomes research on AI governance. That includes policy-focused work, methodological studies, original research, Reviews, Perspectives, and Comments. The editorial signal is clear: trustworthy AI will not emerge from performance metrics alone. It will also depend on rules, oversight, and practical accountability structures that hold up across different countries and communities.
One sentence captures the stakes plainly: a model that cannot be explained, audited, or safely deployed is not trustworthy, no matter how sophisticated it appears in development.
Editorial Leadership and the Call for Diverse Evidence
The collection is led by guest editors Yunlei Li, PhD of Erasmus MC Cancer Institute, University Medical Center Rotterdam, and Yu Liu, PhD of the National University of Singapore. Their appointments place the collection at the intersection of biomedical research, population health, and computational methods, which matches the subject matter of the call.
The journal says it is looking for studies across diverse countries and communities. That emphasis matters in public health because AI tools do not transfer cleanly from one setting to another. Data quality, population structure, language, infrastructure, regulatory expectations, and clinical workflows can all alter how a system behaves once it leaves the environment in which it was built. Research that tests implementation in different settings can therefore be more informative than work confined to a single controlled context.
The collection also leaves room for methodological advances. Validation is a central issue here, because AI tools can appear impressive during development yet fail under external testing or routine use. Evaluation frameworks that examine robustness, interpretability, bias, and safety are likely to be especially relevant to this call. So are studies that examine how people interact with these systems in practice, because trust is not formed in a vacuum. It develops when users can see how a tool reaches its outputs, understand its limits, and verify that it behaves consistently.
Public Health AI Needs More Than Performance Metrics
Public health differs from many other clinical settings because its reach extends beyond individual patients. A miscalibrated system can affect messaging, resource allocation, surveillance priorities, and policy decisions across entire communities. That scale changes the standard of evidence. Technical performance remains necessary, but it is not enough on its own.
The collection’s emphasis on accountability and fairness points to a deeper concern: AI systems can reinforce existing inequities if they are trained on incomplete data, deployed without scrutiny, or interpreted as objective when they are not. Public health institutions increasingly need evidence that a tool can be relied on not just to classify or predict, but to do so transparently and safely in settings where stakes are high and disparities are already present.
The open call invites submissions until 06 August 2027, leaving a long runway for investigators to contribute original studies, analyses of implementation, and policy discussions. Articles will be published in Communications Health as the collection develops. No articles are listed yet, but the editorial intent is unmistakable: trustworthy AI is being treated as a public health question in its own right, not as a peripheral technical concern.