Primary Healthcare Facility Location Optimization in Ado Ekiti Local Government Area Using Machine Learning and Geographic Information Systems

Access to quality healthcare services remains a critical challenge in many developing countries, particularly in rapidly growing urban centres where healthcare facilities are often unevenly distributed. This study assessed the spatial distribution of Primary Healthcare Centres (PHCs) in Ado-Ekiti, Ekiti State, Nigeria, and developed an optimized framework for healthcare facility location planning using Geographic Information Systems (GIS) and Machine Learning techniques. Spatial datasets including healthcare facility coordinates, population density, road networks, land use/land cover, elevation, and slope were integrated within a GIS environment. A total of thirty-four (34) Primary Healthcare Centres were mapped and analyzed using spatial statistical techniques, including Nearest Neighbor Analysis (NNA), while a Random Forest algorithm was employed to identify suitable locations for future healthcare facility development. The NNA result produced a Nearest Neighbor Ratio of 0.729910, a z-score of -3.012856, and a p-value of 0.002588, indicating a statistically significant clustered distribution pattern of healthcare facilities within the study area. The Random Forest model achieved an overall accuracy of 97.617%, with a Root Mean Square Error (RMSE) of 0.002372, Precision of 0.9023, Recall of 0.9045, and F1-score of 0.90511, demonstrating high predictive capability. The findings revealed that healthcare facilities are concentrated within densely populated urban areas, leaving some peripheral communities underserved. The study concludes that integrating GIS and Machine Learning provides an effective decision-support framework for healthcare accessibility assessment and healthcare facility optimization. The approach offers valuable support for evidence-based healthcare planning and sustainable urban development.