Interpretable Machine Learning Models for Predicting Cesarean Delivery in Class III Obese Cohorts
Obese women who undergo unplanned cesarean sections face a higher risk of complications, and these unplanned cesareans often occur after an unsuccessful labor induction. Each year, more American women are at a high risk of such complications, especially in the presence of increased rate of pre-pregn...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10910171/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Obese women who undergo unplanned cesarean sections face a higher risk of complications, and these unplanned cesareans often occur after an unsuccessful labor induction. Each year, more American women are at a high risk of such complications, especially in the presence of increased rate of pre-pregnancy obesity. Although previous research has explored machine learning models to predict cesarean likelihood, there has been limited focus on class III obese women (BMI <inline-formula> <tex-math notation="LaTeX">$\geq 40$ </tex-math></inline-formula>) who have undergone labor induction. In this paper, we study a cohort of class III obese women at our institution who were induced for labor, aiming to identify those at a higher risk for cesarean delivery. We propose a machine learning framework that incorporates a variety of models, including Support Vector Machine, Random Forest, eXtreme Gradient Boosting (XGBoost), K-nearest neighbors, Naïve Bayes, and logistic regression, for predictive analysis. Further, we investigate the significance of individual predictors using SHapley Additive exPlanations (SHAP) method to enhance the interpretability of our models. Moreover, we conduct a comparative analysis of the cohort of nulliparous with combined cohort of nulliparous and multiparous patients. Our comparative analysis shows logistic regression to be the most accurate in predicting the need for cesareans in the nulliparous cohort, while random forest outperformed other models in the combined dataset. |
|---|---|
| ISSN: | 2169-3536 |