Machine Algorithm for Heartbeat Monitoring and Arrhythmia Detection Based on ECG Systems

Area of Science:

• Cardiology
• Biomedical Engineering
• Machine Learning

Background:

• Cardiac arrhythmia involves erratic heartbeats (too slow or fast) due to faulty electrical impulses.
• Serious arrhythmias can lead to sudden cardiac death, necessitating accurate detection.
• Electrocardiogram (ECG) signals (P, QRS, T waves) provide vital data for diagnosing heart conditions.

Purpose of the Study:

• To enhance the reliable perception of life-threatening arrhythmias through advanced ECG signal analysis.
• To introduce a novel machine learning approach for improved cardiac arrhythmia detection.
• To facilitate better diagnosis and timely therapeutic interventions for patients with heart rhythm disorders.

Main Methods:

• Utilized autoregressive (AR) analysis to extract signal features from ECG waveforms.
• Developed a new technique employing two-event-related moving averages (TERMAs) and fractional Fourier transform (FFT) algorithms.
• Implemented cross-database training and testing with enhanced characteristics for a machine learning model.

Main Results:

• AR characteristics effectively separated different ECG signal types in the training dataset.
• Achieved high classification accuracy and reliable heart problem diagnosis using the training data.
• The proposed TERMAs and FFT-based method demonstrated potential for superior ECG signal evaluation.

Conclusions:

• The study highlights the effectiveness of AR signal analysis for ECG-based arrhythmia classification.
• The novel TERMAs and FFT approach offers a promising advancement in ECG signal processing for arrhythmia detection.
• The machine learning model's cross-database capability suggests robust performance across diverse datasets.