Artificial Intelligence in Predicting Childbirth Delivery Modes: A Systematic Review of Model Performance, Validation, and Clinical Integration

Background: Accurate prediction of childbirth delivery modes, including vaginal birth, cesarean section, and Vaginal Birth after Cesarean, is a critical component of obstetric care, with direct implications for maternal safety, neonatal outcomes, and resource allocation. Artificial intelligence offers promising tools for enhancing clinical decision-making in this domain. Objective: This systematic review aimed to identify, synthesize, and critically appraise empirical studies that applied intelligence-based models to predict delivery outcomes, focusing on methodological quality, model performance, and clinical applicability. We systematically searched four databases: PubMed (since ۱۹۹۶), Scopus (since ۲۰۰۴), Web of Science (since ۱۹۹۷), and IEEE Xplore (since ۲۰۰۰) from January ۲۰۱۰ to July ۲۰۲۵. Studies were screened and selected based on predefined inclusion criteria, focusing on AI-based models predicting childbirth delivery modes. Data extraction and quality appraisal were conducted using structured forms and a modified PROBAST framework. Due to methodological heterogeneity, a narrative synthesis was performed. Results: A wide range of machine learning models were employed, including Logistic Regression, Decision Trees, Support Vector Machines, Random Forest, XGBoost, and Deep Neural Networks. Ensemble models such as Random Forest and XGBoost demonstrated superior predictive performance. However, most studies lacked external validation, standardized metrics, and interpretability tools. Only one study incorporated patient preferences, and few addressed ethical or usability concerns. Conclusion: While AI-based models show substantial potential for improving delivery mode prediction, their clinical adoption remains limited by methodological gaps and contextual constraints. Future research should prioritize multicenter validation, explainable frameworks, standardized input variables, and integration into shared decision-making platforms, especially within midwifery-led care. This review provides a reproducible foundation for advancing responsible, patient-centred AI in maternal health care.

کلیدواژه ها:

Artificial Intelligence ، Machine Learning ، Deep Learning ، Mode of Delivery Prediction ، Vaginal Birth after Cesarean ، Cesarean Section