Work

Development of an Artificial Intelligence Model to Predict Mode of Delivery Through Analysis of Labor Partographs

Year: 2026

Type: article

Abstract: Background and Aim: The aim of this study was to develop an artificial intelligence model capable of predicting the mode of delivery (vaginal delivery/cesarean section) using time-series data obtained from labor partographs and to evaluate the contribution of partograph parameters to this prediction. Materials and Methods: A total of 475 cases with complete partograph records among singleton, live pregnancies at 37+0–41+6 weeks of gestation between January 2021 and December 2024 were retrospectively analyzed. Only intrapartum cesarean sections performed for obstetric indications during active labor were included. Patients were divided into two groups: vaginal delivery (n=282) and cesarean delivery (n=193). Cervical dilation, fetal head descent, contraction intensity, latent and active phase durations, and maternal variables were analyzed. A deep learning model based on LSTM was constructed using time-series data, and its performance was evaluated in terms of accuracy, sensitivity, specificity, F1 score, and ROC–AUC. Results: BMI and oxytocin use were higher in the cesarean delivery group. The rates of cervical dilation and fetal head descent, as well as contraction intensity, were significantly higher in the vaginal delivery group (all p<0.001). The LSTM model achieved an accuracy of 87%, sensitivity of 84%, specificity of 83%, and an ROC–AUC of 0.91. Neonatal outcomes were largely similar between the two groups. Conclusion: Time-series analysis of partograph data using artificial intelligence demonstrated high performance in predicting the mode of delivery and has the potential to serve as a clinical decision-support tool during the intrapartum period.

Source: Women’s Health and Multidisciplinary Research

DOI: 10.14744/whmr.2026.35316