Researchers must uphold data equity and health equity principles to improve health outcomes. A special communication from a leading medical journal proposes ten core concepts to produce actionable evidence in health studies. Yiran Wang, PhD, of the Yale School of Public Health, New Haven, was the lead author of this manuscript, published online Friday in JAMA Health Forum. Many populations continue to be underrepresented in health studies, including rural communities, people with disabilities, people who are unhoused and members of low-income regions. Major public health decisions usually rely on large datasets to inform policy-making, and it is increasingly important to address inequities in digital health data to improve outcomes for people across all communities. This special communication highlighted 10 core concepts to consider for equitable research including ideas that bridge gaps between computer science and medicine. The first five concepts fit under the computer science branch: fairness, accountability, transparency, ethics and privacy and confidentiality. Fairness matters for ensuring equal treatment, and accountability holds the artificial intelligence systems responsible for outcomes they produce via algorithms. Transparency is key to understanding results: clear and accessible data to users, the public and anyone interested in the research. Ethics stream across all branches of science, and outline principles of correctness and values. Ethics also prevent harm. Finally, privacy and confidentiality are vital to ensure safety and security for participants in a study. It is important to prevent unauthorized disclosure of health information. The last five concepts are branched under public health: selection bias, representativeness, generalizability, causality and information bias. Selection bias influences the population selected to participate in a study: unseen factors that may influence results. Representativeness is key for a diverse population in a study, where socioeconomic differences are noted, and smaller populations are included. Generalizability means the results can be valid for multiple populations. Causality means labeling the relationship between exposures and outcomes in a study, and this can be a challenge, but it is key to the accuracy of the results of a study. Finally, information bias distorts the measurements used in the study or the variables involved. Information bias can look like missing data or loss of follow-up information from participants. “Together, these steps provide a roadmap for embedding data equity into all phases of public health data science,” the authors concluded. “The implementation of these principles is wrapped in reflexivity, rooted in the specific public health context...Advancing data equity must be accompanied by parallel efforts in information theory and structural changes to empower individuals and communities with informed decision-making around their own and others’ health.” Source: Wang Y, Boyd AE, Rountree L, et al. Ten core concepts for ensuring data equity in public health. JAMA Health Forum. 2026 January 9 (Article in Press). Image Credit: wladimir1804 – stock.adobe.com