Machine Learning Approach to Predict Emergency Cesarean Sections Among Nulliparous Women

Area of Science:

• Obstetrics and Gynecology
• Medical Informatics
• Machine Learning in Healthcare

Background:

• Minimizing cesarean sections in nulliparous women is a key obstetrical goal.
• Identifying risk factors for cesarean delivery is crucial for informed decision-making.
• Machine learning offers a novel approach to predict cesarean delivery risk.

Purpose of the Study:

• To identify predictors of emergency cesarean sections in nulliparous women.
• To evaluate the performance of various machine learning models in predicting cesarean delivery.
• To inform strategies for reducing cesarean birth rates.

Main Methods:

• A retrospective cohort study of 2668 nulliparous women at a tertiary center.
• Analysis of 23 potential risk factors using seven machine learning models.
• Linear regression model identified as the best predictor.

Main Results:

• The overall cesarean section rate was 28.2%.
• Linear regression achieved an AUROC of 0.86, predicting emergency cesarean sections.
• Key predictors included advanced maternal age, education, diabetes, preeclampsia, placenta abruption, hypothyroidism, meconium-stained amniotic fluid, late-term pregnancy, doula support, and prenatal education.

Conclusions:

• Machine learning models, particularly linear regression, show promise in predicting emergency cesarean sections.
• Clinical databases combined with machine learning can enhance prediction accuracy.
• Further prospective research incorporating intrapartum data is needed to refine predictive models.