Enhancing VBAC Prediction with AI-Powered Temporal Dynamics: Integrating Decision Support into a Shared Decision-Making Platform for Intrapartum Care

Background: Taiwan has a high caesarean section (CS) rate, ranging from 37% to 38%. Vaginal Birth After Cesarean (VBAC) offers a potential solution to reduce these rates. However, the prevalence of VBAC remains below 0.5%, primarily due to concerns about risks of adverse maternal and perinatal outcomes. Objectives: This study aims to evaluate the predictive performance of various machine learning (ML) models using pregnancy, labor, and intervention-related features to predict VBAC success and support real-time clinical decision-making during labor. Study Design: This retrospective exploratory study analyzed data collected from a hospital in northern Taiwan between January 2019 and May 2023. Statistical methods included demographic comparisons, feature evaluations, and model performance metrics such as accuracy, precision, recall, F1-score, and area under the curve (AUC). SHapley Additive exPlanations (SHAP) analysis was used to interpret feature importance and labor progression. Results: A comparison between the VBAC Failure group (n=22) and VBAC Success group (n=33), totaling 55 records from 36 pregnant women, revealed significant differences in parity, spontaneous rupture of membranes, cervical dilation (at both 0 cm and 10 cm), and labor progression slope. Models incorporating high-impact features demonstrated superior performance compared to those utilizing only pregnancy-related data. The Random Forest model achieved an accuracy of 94% and an AUC of 0.96 in predicting labor progression. SHAP analysis further identified key predictors across different stages of labor, including pregnancy-related features (body mass index, prior vaginal birth, maternal age), static features (spontaneous rupture of membranes, time since rupture), and dynamic features (cervical dilation and labor slope). Conclusion: This integrative approach, which combines clinical expertise with predictive analytics, provides clinicians with a valuable tool for real-time labor evaluation and decision-making. By offering more accurate predictions of labor progression, particularly in the context of VBAC, this approach has the potential to significantly improve maternal and neonatal outcomes