Ensemble Based Machine Learning Model for Early Detection of Mother's Delivery Mode

Abstract

The mother's mode of delivery greatly impacts the relationship between the newborn baby and the mother, as well as the mother's and baby's health. Currently, the cesarean rate is increasing at an alarming rate. The inability to predict the mother's health status and mode of delivery are mainly responsible for this situation. Support Vector Machine (SVM), Decision Tree, Random Forest (RF), Gradient Boosting Classifier(GBC), Logistic Regression, Gaussian Naive Bayes, Stochastic Gradient Descent, CatBoost (CB), Adaptive Boosting (AB), Gaussian Naïve Bayes, Extreme Gradient Boosting(XGB) are used to predict the mother's mode of delivery. This study also proposed an ensemble machine learning algorithm that stacked the SVC, XGB, and RF together and named the ensemble SVXGBRF. To preprocess the dataset, we use a pipeline that basic preprocessing techniques, data balancing and feature selection. Our proposed SVXGBRF classifiers show 95.52% accuracy, 96% precision, recall, f1 score, and 99% AUC score. SVXGBRF shows its superiority, where most models show an accuracy of less than 90% except RF, GBC, CB, and AB. Eventually, this research could be utilized to develop a decision-support system for reducing the number of cesarean sections by trying to extract insights from complex data patterns.