Arslanoğlu İ., Yıldırım S., Haydaroğlu C.
International Congress on Human-Computer Interaction, Optimization and Robotic Applications (ICHORA), Ankara, Türkiye, 21 - 23 Mayıs 2026, ss.1-8, (Tam Metin Bildiri)
-
Yayın Türü:
Bildiri / Tam Metin Bildiri
-
Doi Numarası:
10.1109/ichora69329.2026.11536993
-
Basıldığı Şehir:
Ankara
-
Basıldığı Ülke:
Türkiye
-
Sayfa Sayıları:
ss.1-8
-
Karadeniz Teknik Üniversitesi Adresli:
Evet
Özet
Electrocardiogram (ECG) signal classification plays a critical role in the early and reliable detection of cardiovascular diseases. However, noise contamination, high variability, and inter-class overlap significantly limit the performance of conventional methods, particularly in multi-class arrhythmia detection. To address these challenges, this study proposes a hybrid machine learning (ML) framework integrating Fuzzy C-Means (FCM) clustering for enhanced feature representation. The experiments are conducted on the PhysioNet MIT-BIH Arrhythmia dataset, consisting of approximately 110,000 samples, divided into 80% training and 20% testing sets. ECG signals are preprocessed using a 0.5−40Hz band-pass filter, and heartbeat segments are extracted via R-peak detection. Each sample is represented by time-series data points, from which statistical and structural features (mean, standard deviation, maximum, minimum, range, and energy) are derived. The obtained fuzzy membership values are fused with the original features to construct a hybrid feature space. Random Forest (RF), Support Vector Machine (SVM), and XGBoost classifiers are then trained using both classical and FCM-enhanced feature sets. Experimental results demonstrate that the proposed FCM-based feature fusion improves classification performance, particularly for ensemble learning models. The RF model achieves the highest performance (accuracy ≈0.9999), while XGBoost shows notable improvement compared to the classical feature-based approach (accuracy ≈0.9996). The SVM model maintains stable and consistent performance across both scenarios (accuracy ≈0.996). Overall, the findings indicate that incorporating fuzzy membership information enhances feature representation by effectively modeling uncertainty and class overlap, leading to more robust and reliable multi-class ECG classification.