Explainable machine learning analysis of factors associated with skilled birth attendance in Burkina Faso

Abstract Skilled birth attendance (SBA) is essential for reducing maternal and neonatal mortality, but its coverage varies Burkina Faso. This study used explainable machine learning to predict SBA, identify key predictors, and examine geographic and urban–rural descriptive inequalities. This cross-sectional study analyzed 5,111 women aged 15–49 years from the 2021 Burkina Faso Demographic and Health Survey (BF-DHS). Features were selected using Boruta feature selection, and SMOTE used for class imbalance. Five machine learning models used in this study such as Random Forest (RF), Decision Tree (DT), K-Nearest Neighbors (KNN), Logistic Regression (LR), and Support Vector Machine (SVM) were trained. Model performance was evaluated using accuracy, precision, recall, F1-score, Matthew’s correlation coefficient (MCC), Cohen’s kappa, and AUROC. SHapley Additive exPlanations (SHAP) interpreted model predictions, while decision curve analysis (DCA) and spatial mapping evaluated clinical utility and geographic variation. Among five models, RF achieved the highest discrimination, with moderate performance (AUROC = 0.71). The key predictors included province, antenatal care visits ≥ 4, maternal age at first birth ≥ 20 years, age at first sexual intercourse ≥ 18 years, sexual activity, household wealth, and religion. Descriptive spatial mapping showed the higher predicted SBA probabilities in central and western provinces and lower probabilities in Sahel, Sud-Ouest, and Est, with greater inequalities in rural areas rural areas. These results highlight that interpretable machine learning combined with spatial analysis can identify predictors of SBA and capture subnational inequalities, offering reproducible, data-driven insights for maternal health research.