Submission ID 130366
| Issue/Objective | Despite significant advances in global health, inequities in access to quality healthcare and health outcomes persist, particularly among underserved and marginalized populations. Structural barriers-including poverty, limited health system capacity, geographic isolation, and fragmented governance-continue to widen disparities, especially in low- and middle-income settings. At the same time, the rapid expansion of digital health and data systems presents new opportunities to address these inequities; however, translating data into actionable, equity-focused interventions remains a critical challenge. Artificial intelligence (AI) and data-driven approaches have shown promise in improving disease surveillance, resource allocation, and decision-making. Yet, their application in advancing health equity is often limited by gaps in implementation, contextual adaptation, and alignment with community needs. There is an urgent need to move beyond theoretical models toward practical, scalable solutions that integrate AI into real-world public health systems in ways that are equitable, inclusive, and responsive to local contexts. This study aims to explore how AI-driven tools can be leveraged to improve equitable health outcomes in underserved populations by translating data insights into actionable public health interventions. Specifically, the objective is to demonstrate how data-informed approaches can support decision-making, enhance resource targeting, and strengthen health system responsiveness, thereby contributing to more just and effective health outcomes. This work aligns with the CCGH 2026 theme, "Collectively reclaiming global health equity," by emphasizing the transition from evidence to action and highlighting the role of innovative, equity-driven solutions in addressing persistent global health disparities. |
|---|---|
| Methodology/Approach | This study employs a mixed methods implementation research design conducted in Nigeria, focusing on underserved rural and peri urban communities where structural inequities continue to limit access to essential healthcare services. The approach integrates quantitative data analysis with contextual assessment to ensure that findings are relevant, actionable, and responsive to the needs of marginalized populations. Quantitative data are obtained from Demographic and Health Surveys, District Health Information System version 2, and national health surveillance datasets. These datasets include indicators of healthcare access, disease burden, and key social determinants of health such as income, education, and geographic location. Machine learning models, including random forest and logistic regression, are applied to identify high risk populations and geographic areas experiencing disparities in health outcomes and service access. Model outputs are translated into risk stratification maps and equity focused priority indices to support targeted resource allocation. These tools are designed to assist health planners in identifying communities that are systematically underserved and at increased risk of poor health outcomes. To ensure contextual relevance, a qualitative component is incorporated through structured review of policy documents, implementation reports, and evidence from similar low resource settings. This enables assessment of feasibility, scalability, and alignment with existing health system structures and priorities. The implementation strategy focuses on developing practical decision support tools, including simplified dashboards and prioritization frameworks, that can be integrated into routine public health planning processes. These tools are intended to strengthen equitable service delivery and improve responsiveness to vulnerable populations. The study is conducted over six months, including data acquisition, model development, validation, and translation into actionable outputs. The overall approach emphasizes bridging the gap between data and action, ensuring that technological innovations contribute to equitable and people centered health system improvements. |
| Results | The analysis identified clear disparities in healthcare access and health outcomes across rural and peri urban communities, strongly associated with socioeconomic status, geographic location, and health system capacity. Machine learning models demonstrated good predictive performance, with area under the receiver operating characteristic curve values exceeding 0.80, indicating reliable identification of high risk populations and underserved areas. Risk stratification analysis revealed that communities with lower income levels, limited health facility coverage, and reduced access to transportation had significantly higher predicted risks of poor health outcomes, including delayed care seeking and increased disease burden. These findings highlight the compounding effect of structural inequities on health access and outcomes. The generated priority indices and mapping outputs enabled identification of specific high need areas where targeted interventions such as mobile health services, outreach programs, and resource redistribution could have the greatest impact. Scenario based analysis showed that applying data informed targeting could improve efficiency of resource allocation by directing limited health resources to the most vulnerable populations. The implementation component demonstrated that translating model outputs into simplified decision support tools is feasible and can enhance local level planning. The use of interpretable dashboards and prioritization frameworks improved the ability to visualize disparities and support evidence informed decision making. Overall, the findings indicate that artificial intelligence driven approaches can effectively identify inequities and support actionable strategies to improve equitable health outcomes. Importantly, the study demonstrates that when combined with context sensitive implementation, data driven tools can contribute to more responsive and inclusive health systems in resource constrained settings. |
| Discussion/Conclusion | This study demonstrates the potential of artificial intelligence driven approaches to move beyond data analysis toward actionable, equity focused public health interventions. By identifying underserved populations and translating insights into practical decision support tools, the findings highlight how data can be leveraged to address persistent structural inequities in healthcare access and outcomes. The implications for policy and practice are significant. Integrating data driven prioritization into routine health system planning can support more efficient and equitable allocation of limited resources, particularly in low resource settings. Policymakers and health system leaders can use these tools to identify high need communities, guide targeted interventions such as outreach services and mobile care delivery, and strengthen responsiveness to vulnerable populations. Importantly, the approach emphasizes the need to align technological innovation with local context, governance structures, and community needs to ensure equitable impact. In terms of scalability, the methodology relies on widely available datasets and adaptable analytical frameworks, making it feasible for replication across different regions and health systems. With appropriate capacity building and stakeholder engagement, these tools can be integrated into existing health information systems to support sustainable and context sensitive implementation at scale. This work contributes directly to the CCGH 2026 theme of "Collectively reclaiming global health equity" by demonstrating how evidence informed, technology enabled solutions can support concrete action in addressing health disparities. In alignment with Sub theme 3, it bridges the gap between research and practice by transforming data into actionable strategies that promote accountability, strengthen health systems, and advance equitable health outcomes in a fragmented and resource constrained global landscape. |
| Presenters and Affiliations | Kelvin Ovabor The University of Alabama Kelvin Ovabor The University of Alabama Adaeze Nweke The University of Alabama |
