Predicting Air Conditioner Ownership in Low- and Middle-Income Countries using Geospatial Machine Learning Methods - Mapping Cooling Inequality

Abstract Background: Rising global temperatures are intensifying exposure to extreme heat, yet access to air conditioning (AC) remains unequal which leaves vulnerable populations without protection against the heat. Household survey data on AC ownership is scarce in low- and middle-income countries limiting the ability to identify areas of cooling gaps. Methods: In this study, we develop a geospatial machine learning framework that integrates household data from the Demographic and Health Surveys (DHS) program with high resolution climatic, socioeconomic and demographic covariates to generate a 5km resolution prediction of AC ownership across low- and middle-income countries in Africa and Asia for the year 2023. Results: The resulting map reveals strong spatial inequalities with high predicted prevalence in wealthier and urban regions, and near zero prevalence across much of Central and West Africa and rural areas. These patterns reflect the adaptation cooling gap where those most exposed to extreme heat are least able to adapt through access to cooling. Conclusion: By producing the first continental-scale, high resolution map of AC ownership, this study lays the foundation for designing and targeting equitable cooling strategies, support heat-health planning particularly in communities most affected and informing sustainable energy policy.