Identifying Determinants of Cesarean Births: A Machine Learning Framework for Maternal Health Research

Abstract Cesarean section is a vital obstetrical procedure that is meant to minimize the dangers to the mother and the baby. Nevertheless, the relentless increases in cesarean rates and race and ethnic disparities are a significant public health issue. This is a study of cesarean birth determinants based on the 2022 U.S. National Vital Statistics System (NVSS) natality dataset, which consists of more than 3.5 million live singleton institutional births. We compared three supervised machine learning algorithms, Random Forest, Gradient Boosting, and Extreme Gradient Boosting (XGBoost), to determine risk of cesarean delivery using 41 demographic, clinical, and prenatal predictors. Of them, XGBoost model showed better predictive behavior (accuracy = 79.1%, F1 = 0.63, ROC-AUC = 0.85). SHAP (SHapley Additive Explanations) values were calculated to understand model outputs and found that the major contributors included labor interventions, maternal age, and fetal presentation. Results demonstrate the future of machine learning in predicting high-risk obstetric outcomes and guide the process of minimizing unnecessary cesareans particularly among underserved groups. The study provides an interpretable, scalable method of maternal risk prediction that can be used in clinical decision-making.