Enhancing Obstetric Decision-Making With AI: A Systematic Review of AI Models for Predicting Mode of Delivery
Selma Mohammed Abdelgadir Elhabeeb1, Sulafa Hassan Mahmoud Ali2, Marwa Mohamed Ahmed Elkhidir Babikir3
1Obstetrics and Gynecology, Najran Armed Forces Hospital, Ministry of Defense Health Services, Najran, SAU.
Cureus
|June 9, 2025
Summary
Artificial intelligence (AI) models show promise in predicting delivery mode, aiding obstetric decisions and reducing cesarean sections. Ensemble and real-time AI models offer the highest accuracy, but external validation and interpretability are key for clinical use.
Area of Science:
- Obstetrics and Gynecology
- Medical Artificial Intelligence
- Predictive Analytics in Healthcare
Background:
- Accurate prediction of delivery mode is crucial for maternal and neonatal outcomes.
- Unnecessary cesarean sections pose risks and increase healthcare costs.
- Artificial intelligence (AI) offers potential for improved obstetric decision-making.
Purpose of the Study:
- To systematically review and synthesize evidence on AI models for predicting delivery mode.
- To compare the performance and clinical applicability of various AI models.
- To identify key factors influencing AI model accuracy and translation to practice.
Main Methods:
- Comprehensive literature search for AI-based predictive models of delivery mode.
- Analysis of 17 studies using diverse AI techniques (e.g., Random Forest, neural networks, ensemble methods).
- Systematic extraction and comparison of study characteristics, input variables, performance metrics, and validation methods.
Main Results:
- AI models demonstrated good to excellent predictive performance (AUC values).
- Ensemble models and real-time intrapartum data significantly improved accuracy.
- Maternal age, parity, BMI, prior cesarean, and clinical data were common predictors.
- Advanced AI methods often outperformed logistic regression, with simpler models valued for interpretability.
Conclusions:
- AI holds substantial potential for enhancing delivery mode prediction and supporting clinical decisions.
- Ensemble and dynamic AI models show the highest predictive power.
- External validation, model transparency, and clinical integration remain critical challenges.
