Identification and classification of arrhythmic heartbeats from electrocardiogram signals using feature induced...

Area of Science:

• Cardiology
• Biomedical Engineering
• Machine Learning

Background:

• Cardiovascular diseases are a leading cause of death globally.
• Early detection of arrhythmic heartbeats is crucial for mitigating risks.
• Automated systems can aid in timely diagnosis and treatment.

Purpose of the Study:

• To develop a computer-aided diagnostic (CAD) approach for automated identification and classification of arrhythmic heartbeats.
• To enhance the accuracy of arrhythmia detection using machine learning techniques.
• To improve the clinical utility of electrocardiogram (ECG) signal analysis.

Main Methods:

• Utilized the MIT-BIH Arrhythmia database for training and testing.
• Applied discrete wavelet transform for ECG signal pre-processing.
• Extracted statistical, temporal, and spectral features.
• Employed random forest for feature selection and an optimal extreme gradient boosting (O-XGBoost) classifier.
• Optimized hyperparameters and used synthetic minority over-sampling technique for dataset balancing.

Main Results:

• Achieved remarkable performance in classifying arrhythmic heartbeats.
• Demonstrated superior results compared to existing state-of-the-art methods.
• The O-XGBoost classifier, with optimized parameters, showed high accuracy.

Conclusions:

• The proposed CAD system effectively classifies arrhythmic heartbeats using ECG signals.
• The approach offers a robust solution for cardiovascular disease detection.
• The model is adaptable for implementation in various computer-aided diagnostic systems.