AUTHOR=Kassie Sisay Yitayih , Abuhay Abebe Solomon , Wondirad Mekdes , Fantaw Muket Samrawit , Melke Ayantu , Chereka Alex Ayenew , Ambachew Shibabaw Adamu , Dubale Abiy Tasew , Damtie Yitayish , Ngusie Habtamu Setegn , Walle Agmasie Damtew 

TITLE=Predictors of community-based health insurance enrollment among reproductive-age women in Ethiopia based on the EDHS 2019 dataset: a study using SHAP analysis technique, 2024

JOURNAL=Frontiers in Public Health

VOLUME=Volume 13 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/public-health/articles/10.3389/fpubh.2025.1448055

DOI=10.3389/fpubh.2025.1448055

ISSN=2296-2565

ABSTRACT=BackgroundOut-of-pocket payments for health services can lead to health catastrophes and decreased service utilization. To address this issue, community-based health insurance has emerged as a strategy to provide financial protection against the costs of poor health. Despite the efforts made by the government of Ethiopia, enrollment rates have not reached the potential beneficiaries. Therefore, this study aimed to predict and identify the factors influencing community-based health insurance enrollment among reproductive-age women using SHapley Additive exPlanations (SHAP) analysis techniques.MethodThe study was conducted using the recent Demographic Health Survey 2019 dataset. Eight machine learning algorithm classifiers were applied to a total weighted sample of 9,013 reproductive-age women and evaluated using performance metrics to predict community-based health insurance enrollment with Python 3.12.2 software, utilizing the Anaconda extension. Additionally, SHAP analysis was used to identify the key predictors of community-based health insurance enrollment and the disproportionate impact of certain variables on others.ResultThe random forest was the most effective predictive model, achieving an accuracy of 91.64% and an area under the curve of 0.885. The SHAP analysis, based on this superior random forest model, indicated that residence, wealth, the age of the household head, the husband’s education level, media exposure, family size, and the number of children under five were the most influential factors affecting enrollment in community-based health insurance.ConclusionThis study highlights the significance of machine learning in predicting community-based health insurance enrollment and identifying the factors that hinder it. Residence, wealth status, and the age of the household head were identified as the primary predictors. The findings of this research indicate that sociodemographic, sociocultural, and economic factors should be considered when developing and implementing health policies aimed at increasing enrollment among reproductive-age women in Ethiopia, particularly in rural areas, as these factors significantly impact low enrollment levels.