Akademik Veri Yönetim Sistemi
2025 Innovations in Intelligent Systems and Applications Conference (ASYU), Bursa, Türkiye, Bursa, Türkiye, 10 Eylül - 12 Kasım 2025, ss.1-6, (Tam Metin Bildiri)
Preeclampsia is a hypertensive disorder that typically arises at the end of the 2nd trimester or the beginning of the 3rd trimester of pregnancy and is a major cause of maternal-fetal morbidity and mortality. In this study, four different machine learning algorithms (CatBoost, Random Forest, LightGBM, and Multilayer Perceptron) and a hybrid model created by combining these models with the soft voting ensemble method are comparatively evaluated for early prediction of preeclampsia risk. In the experimental studies performed on two distinct datasets, the Recursive Feature Elimination (RFE) method was implemented for feature selection, and the class imbalance problem was addressed with the SMOTE algorithm. The Soft Voting Ensemble Model (SVEM) demonstrated superior performance, with 92.50% accuracy and 97.63% Area Under the Curve-Receiver Operating Characteristic (AUC-ROC) values in the factors for Preeclampsia dataset and 92.16% accuracy and 93.35% AUC-ROC values in the Maternal Health Risk dataset. The Ensemble approach reduced the rates of false positive and false negative results by ensuring balanced classification success between positive and negative classes. The findings indicate that the proposed approach can serve as a reliable instrument within the framework of clinical decision support systems.