Applying machine-learning models to successful vaginal birth after two cesareans.

Area of Science:

• Obstetrics and Gynecology
• Reproductive Medicine
• Maternal-Fetal Medicine

Background:

• Vaginal birth after two cesarean deliveries (VBAC2) is a growing consideration for women.
• While VBAC2 is feasible, especially in grand multiparous women, predictors of success and risk reduction are not fully understood.
• Identifying factors for safe VBAC2 is crucial for informed delivery decisions.

Purpose of the Study:

• To identify predictors of successful vaginal birth after two cesarean deliveries (VBAC2).
• To evaluate the safety and success rates of VBAC2 in a large cohort.
• To apply machine learning models for predicting VBAC2 outcomes.

Main Methods:

• A retrospective observational study of 541 women with two prior cesarean deliveries attempting VBAC2.
• Machine learning models (XGBoost Classifier) were used to predict VBAC2 success and safety.
• Primary outcome: successful VBAC2 without uterine rupture; secondary outcomes: Apgar score, NICU admission, postpartum transfusion.

Main Results:

• 78.9% of women achieved a successful VBAC2, with a uterine rupture rate of 2.0%.
• The XGBoost Classifier demonstrated 83% sensitivity in predicting successful VBAC2.
• Significant predictors of VBAC2 success included time since last cesarean, prior vaginal births, and cervical dilation.

Conclusions:

• Grand multiparous women exhibit higher VBAC2 success rates with low uterine rupture risk.
• Machine learning models effectively predict the likelihood of safe VBAC2, aiding clinical decision-making.
• Prior vaginal birth and cervical dilation are key factors influencing VBAC2 success.