A Novel Active Learning Technique for Fetal Health Classification Based on XGBoost Classifier

This project introduces a novel active learning approach for fetal health classification using Cardiotocogram (CTG) data. CTG exams are cost-effective and non-invasive, making them especially valuable in low-resource healthcare environments. The dataset, comprising 2,126 CTG records labeled by expert obstetricians into three categories—Normal, Suspect, and Pathological—serves as the foundation for developing a robust machine learning model. The proposed method leverages ensemble algorithms including XGBoost, Random Forest, CatBoost, and a Stacking Classifier to enhance predictive accuracy. By incorporating active learning, the system efficiently identifies the most informative data points, thereby improving model performance while reducing labeling efforts. This intelligent classification system aims to support timely and informed medical decisions, ultimately contributing to reduced fetal and maternal health risks. The approach demonstrates the potential of data-driven tools in augmenting clinical assessments and improving outcomes in prenatal care.

Keywords: Fetal health, Cardiotocogram, Active learning, Machine learning, XGBoost, Random Forest, CatBoost, Stacking classifier, Maternal health, Prenatal care