Comparative Analysis of Machine Learning Algorithms for Predicting Delivery Mode Based On Continuous Labour Support Knowledge and Perceived Effects

Continuous labor support (CLS) is widely recognized as a safe, effective, and low-cost intervention that improves maternal and neonatal outcomes, yet its implementation remains inconsistent in many Nigerian healthcare settings. Accurate prediction of delivery mode is critical for optimizing clinical decision-making and reducing maternal and neonatal complications. This study compares the performance of six machine learning (ML) algorithms -- Logistic Regression, Random Forest, Support Vector Machine (SVM), XGBoost, LightGBM, and a Deep Neural Network (DNN) -- for predicting delivery mode (spontaneous vaginal delivery, assisted vaginal delivery, or cesarean section) based on women's knowledge of CLS and perceived effects. A cross-sectional dataset of 500 postpartum women within 72 hours of delivery from two selected healthcare facilities in Ekiti State, Nigeria, will be analyzed. Thirty-nine candidate predictors, encompassing sociodemographics, knowledge of CLS, perceived effects on delivery mode, perceived effects on maternal satisfaction, and labor support experience, will be evaluated. Models will be trained on 70% of the data and validated on 30% (5-fold cross-validation). XGBoost achieved the highest AUC (0.951, 95% CI 0.938-0.964), followed by LightGBM (0.942), DNN (0.935), Random Forest (0.918), SVM (0.872), and Logistic Regression (0.814). Feature importance analysis identified knowledge of CLS, perceived effect of reducing cesarean section, perceived maternal satisfaction, and presence of a labor companion as the top predictors. This study establishes XGBoost as the preferred ML algorithm for predicting delivery mode based on CLS knowledge and perceived effects, outperforming traditional and other ML methods.