A HUMAN-CENTERED MACHINE LEARNING DECISION SUPPORT MODEL FOR IDENTIFYING UNREPORTED HYPERTENSION

Résumé fourni par la source

Large numbers of adults live with elevated blood pressure without realizing it, often until complications emerge. This study develops a human-centered machine learning (ML) decision support framework that estimates the likelihood of unreported hypertension by fusing routinely captured clinical records with demographic, behavioral, psychosocial, and wearable-derivedsignals. The framework translates model outputs into clear, context-aware guidance that nudges users toward timely screening and preventive action. Two brief case vignettes—from rural and urban settings—illustrate how alerts can surface hidden risk and reinforce healthy behaviors. In comparative testing, ensemble learners, particularly gradient boosting, delivered stronger discrimination than baseline models while preserving operational practicality. The approach is positioned to complement public-health efforts and clinical care pathways by offering a feasible route to earlier detection and more equitable cardiovascular outcomes. The framework's feasibility is grounded in prior Nigerian research on IoT-based health monitoring and cloud-based e-Health security, demonstrating local capacity for implementation.