Ensemble classifier fostered detection of arrhythmia using ECG data.

Area of Science:

• Cardiology
• Biomedical Engineering
• Machine Learning

Background:

• Electrocardiogram (ECG) is crucial for diagnosing heart conditions like arrhythmia.
• Automatic ECG analysis aids cardiologists in patient monitoring and diagnosis.
• Accurate arrhythmia classification is essential for effective cardiac patient management.

Purpose of the Study:

• To propose an ensemble classifier for accurate arrhythmia detection using ECG signals.
• To evaluate the performance of the proposed method against existing arrhythmia classification models.
• To enhance the diagnostic capabilities of cardiac patient monitoring systems.

Main Methods:

• Utilized the MIT-BIH arrhythmia dataset for training and testing.
• Pre-processed ECG data using Python in a Jupyter Notebook environment.
• Applied Residual Exemplars Local Binary Pattern for feature extraction.
• Employed an ensemble of Support Vector Machines (SVM), Naive Bayes (NB), and Random Forest (RF) classifiers.

Main Results:

• The proposed AD-Ensemble SVM-NB-RF method demonstrated superior performance.
• Achieved significant improvements in accuracy, Area Under the Curve (AUC), and F-Measure compared to deep learning and other ensemble models.
• Outperformed existing models by up to 44.57% in accuracy, 3.33% in AUC, and 23.05% in F-Measure.

Conclusions:

• The AD-Ensemble SVM-NB-RF method offers a highly accurate and effective approach for arrhythmia detection.
• This method provides a valuable tool for improving cardiac patient monitoring and diagnosis.
• The ensemble learning strategy combined with effective feature extraction enhances ECG signal classification performance.