TITLE:
Socio-Behavioral and Spatial Determinants of HIV/AIDS Incidence in Ghana: An Ecological Cross-Sectional Study with Explainable Machine Learning
AUTHORS:
Valentine Golden Ghanem
KEYWORDS:
HIV/AIDS, Ghana, Spatial Epidemiology, Sociobehavioral Determinants, Explainable Machine Learning, Public Health Disparities, ART Coverage, Stigma, Education, SHAP
JOURNAL NAME:
Advances in Infectious Diseases,
Vol.15 No.4,
October
11,
2025
ABSTRACT: Although there has been a drop in cases nationally, Ghana still suffers from severe subnational gaps in HIV/AIDS incidence because of educational inequities, stigma, HIV awareness, and access to antiretroviral therapy (ART). The inequities that lie beneath the surface are frequently concealed by national averages, which reduce the effectiveness of public health interventions. High-incidence and regionally vulnerable areas were identified through spatial clustering and choropleth mapping using a unified, regionally defined dataset from 2000 to 2022 that incorporated socio-behavioral and health infrastructure factors. The Random Forest and XGBoost models, explained through SHAP values, PDPs, and counterfactuals, showed the highest HIV incidence in the Greater Accra, Ashanti, and Central regions, strongly associated with urbanization, stigma, and limited ART access. The analyses further indicated that incidence declined with greater educational access and ART coverage, supporting the need for policies that curb stigma, address regional inequities, and improve access.