Exploring Machine Learning Algorithms for Predicting Early Antenatal Care Initiation at First Trimester among Reproductive Women in Nigeria

Abstract

Background Early antenatal care (ANC) initiation during the first trimester is crucial for maternal and child health outcomes. However, in Nigeria, early ANC uptake remains low due to socioeconomic and cultural barriers. Traditional statistical models used to predict ANC initiation often fail to capture the complex nonlinear interactions between predictors. This study applies machine learning (ML) algorithms to predict early ANC initiation using data from Nigeria’s 2018 Demographic and Health Survey (NDHS). Methods This cross-sectional study utilized NDHS 2018 data and six ML algorithms: Logistic Regression (LR), Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Decision Tree, Random Forest (RF), and XGBoost. The synthetic minority oversampling technique (SMOTE) was used to address data imbalance. The models were evaluated using metrics, such as accuracy, precision, recall, and AUROC values. Feature importance was assessed using the permutation importance and Gini impurity methods. Results The Random Forest model demonstrated the best performance, achieving the highest accuracy (77%), precision (75%), recall (80%), and AUROC value of 0.77. This model effectively identified key predictors of early ANC, including education level, wealth index, and place of delivery, with educated and wealthier women being more likely to seek early care. Feature importance analysis revealed significant sociodemographic and geographic disparities, highlighting the critical role of economic resources and healthcare accessibility in influencing ANC behavior. Conclusion ML models, particularly Random Forest, provide an effective approach for predicting early ANC initiation, offering comprehensive insights beyond traditional methods. These findings underscore the need for targeted interventions focusing on education, economic empowerment, and healthcare access to enhance early ANC uptake among Nigerian women.