ECG Heartbeat Classification Using Machine Learning and Metaheuristic Optimization for Smart Healthcare Systems.

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Healthcare
• Cardiovascular Disease Diagnostics

Background:

• Early diagnosis of cardiac arrhythmias from ECG is crucial for cardiovascular health monitoring.
• Traditional machine learning (ML) classifiers struggle with ECG's nonlinearity and low amplitude, limiting accuracy.
• High-dimensional ECG data features and complex interrelationships pose challenges for existing ML models.

Purpose of the Study:

• To develop an automatic arrhythmia classification system for improved smart healthcare.
• To enhance the performance of ML classifiers by optimizing their search parameters.
• To address the limitations of traditional ML in analyzing complex ECG data.

Main Methods:

• An approach integrating a metaheuristic optimization (MHO) algorithm with supervised ML classifiers (SVM, kNN, GBDT, RF).
• ECG signal preprocessing, feature extraction, and classification steps.
• MHO algorithm used to optimize learning parameters of the selected ML classifiers.

Main Results:

• Significant performance improvement across all tested ML classifiers after MHO integration.
• Achieved an average ECG arrhythmia classification accuracy of 99.92% and sensitivity of 99.81%.
• Outperformed existing state-of-the-art methods on MIT-BIH, EDB, and INCART databases.

Conclusions:

• The proposed MHO-integrated ML approach effectively enhances ECG arrhythmia classification.
• This method offers a robust solution for smart healthcare systems monitoring cardiovascular health.
• The approach demonstrates superior accuracy and sensitivity, paving the way for advanced diagnostic tools.