Arrhythmia Detection Using WPD with Bagging and Boosting Ensemble Machine Learning Methods
Subasi, Muhammed Enes
Subasi, Muhammed Enes
Type
Supervisor
Date
2026-04
Collections
Research Projects
Organizational Units
Journal Issue
Abstract
Arrhythmias (ARRs), abnormal heart rhythms, pose significant risks to cardiovascular health and require prompt detection and treatment to mitigate adverse outcomes. Recent advancements in signal processing and machine learning have led to the creation of effective ARR detection systems. This chapter introduces a novel method for detecting ARRs using wavelet packet decomposition (WPD) combined with ensemble methods to enhance diagnostic accuracy and reliability. The approach includes several key steps: acquiring and preprocessing ECG signals to remove noise and artifacts, applying WPD to decompose the signals into multiple frequency sub-bands, and extracting features that characterize different ARRs. Ensemble methods, including bagging, AdaBoost, and MultiBoosting, are then used to combine multiple classifiers into a unified decision-making framework. Extensive experiments using the MIT-BIH Arrhythmia Database, which contains various ARR types and severity levels, demonstrate the efficacy of the proposed detection system. Performance metrics such as accuracy, F-measure, area under the curve, and Kappa evaluate the system’s ability to distinguish between normal and abnormal heart rhythms. The results show that combining WPD and ensemble methods significantly improves performance compared to individual classifiers and traditional signal processing techniques. The integration of WPD and ensemble methods enhances the accuracy and reliability of ARR detection systems. This combined approach effectively utilizes the strengths of diverse classifiers, resulting in better detection accuracy, noise robustness, and generalization across different datasets and ARR types.
Department
Publisher
Sponsor
Effat University
Copyright
Book title
Time-Frequency Analysis in Biomedical Engineering