Introduction: A Data-Driven Shift in Public Health
The public health landscape is changing rapidly as vast datasets—from electronic health records to wearables, imaging, and social media—inform decision‑making. This convergence of data science, AI, and multimodal data holds the promise of accelerating health improvements, yet it also risks widening disparities if equity is not embedded in data practices. The Yale School of Public Health conference on data equity and health innovation (April 2024) underscored the imperative to align data collection, analysis, and insight generation with principles of fairness, human rights, and community voice.
Three Thematic Lenses: SDOH, AI, and Community Data
Discussions during the conference centered on three interrelated themes: how to depict social determinants of health (SDOH) in data; the impact of artificial intelligence (AI) on health data equity; and community-based models for data. Together, these lenses provide a framework for turning data into equitable health action rather than just more information.
Depicting SDOH in Data
SDOH—factors such as economic stability, education access, healthcare availability, neighborhood conditions, and social context—explain up to a substantial portion of health outcomes. The challenge is to extend data collection beyond traditional clinical sources. Tools like CMS’s Health-Related Social Needs screenings and expansion of ICD-10 Z-codes to capture social services are steps forward, yet most SDOH data remain outside formal health systems. Innovations in natural language processing, including large language models to extract SDOH signals from EHR narratives, show promise but require careful attention to privacy, consent, and purpose.
AI and Health Data Equity
AI offers transformative potential but also introduces risks of bias and unequal benefits. The conference highlighted strategies to prevent amplification of disparities, including community engagement in AI development, transparency standards, and explicit bias assessment. Frameworks such as ACCESS AI and health AI assurance guides were cited as tools to ensure that AI health applications reflect diverse populations and are auditable, explainable, and accountable.
Community-Based Data Models
Community engagement is central to data equity. Participatory approaches—where communities define priorities, co-create data collection methods, and co-own the use of insights—help ensure relevance and trust. The session identified three pillars: defining community in context, promoting equity through collaboration, and enhancing representation by blending quantitative and qualitative data. Decentralized data models, in which communities retain control over their data, emerged as a promising path to sovereignty and more ethically grounded health innovation.
Key Insights and Practical Recommendations
From the roundtables, several cross-cutting recommendations emerged for policymakers and practitioners seeking to embed data equity into health innovation:
- Enable big data and interoperability while safeguarding privacy and consent, so SDOH data can be linked with health and social services across settings.
- Include diverse, nontechnical voices in AI governance to capture sociotechnical perspectives and build trust with communities.
- Foster collaboration among universities, industry, and local systems to steward data responsibly and sustain equity-focused innovations.
- Modernize HIPAA and create AI-specific guidelines that address transparency, bias mitigation, and data equity.
- Develop new conceptual frameworks that explicitly connect data equity with health outcomes, guiding ethical AI deployment and inclusive research.
These recommendations acknowledge fiscal and infrastructural constraints, especially amid shifts in Medicaid funding and public health data system unwinding. They call for scalable, collaborative approaches—grounded in community engagement and accountable governance—that can adapt to resource variability while advancing equitable health outcomes.
Conclusion: Turning Data into Fair Health Outcomes
The conference reaffirmed that data equity is not a peripheral consideration but a central pillar of future public health. Real progress hinges on breaking data silos, enhancing interoperability, and ensuring AI applications are transparent, participatory, and rooted in diverse community experiences. By prioritizing inclusive data practices, policymakers and practitioners can align health innovation with the goal of improving population health for all, not just a subset of communities.
