Analysis of Fetal Health Classification Using Machine Learning Models During Childbirth

Abstract

Accurate monitoring of fetal health during labor is essential for preventing adverse neonatal outcomes such as fetal hypoxia, neurological injury, and perinatal mortality. Cardiotocography (CTG) is widely used to monitor fetal heart rate and uterine contractions; however, its interpretation often involves subjectivity and interobserver variability among clinicians. Recent advances in machine learning offer promising approaches for improving the accuracy and objectivity of fetal health classification using CTG data. This study aims to analyze the effectiveness of machine learning models in classifying fetal health conditions during labor. The research employed a quantitative approach using a machine learning–based classification framework with CTG datasets containing fetal heart rate and uterine contraction features. Data preprocessing included cleaning, normalization, and feature selection, followed by model training using Support Vector Machine (SVM), Random Forest (RF), and Extreme Gradient Boosting (XGBoost). Model performance was evaluated using accuracy, precision, recall, F1-score, and ROC-AUC metrics. The results indicate that machine learning models achieve high classification performance, with XGBoost producing the best accuracy and discrimination capability among the tested algorithms. These findings demonstrate that machine learning approaches can effectively analyze complex physiological patterns in CTG data and support clinical decision making during labor. However, challenges related to dataset diversity, clinical standardization, and external validation remain important considerations for future clinical implementation.