An Efficient Cardiac Arrhythmia Onset Detection Technique Using a Novel Feature Rank Score Algorithm.

Area of Science:

• Cardiovascular physiology
• Biomedical signal processing
• Medical diagnostics

Background:

• Cardiovascular blood flow abnormalities, particularly left-ventricular arrhythmias (LVA), are significant health concerns.
• Electrocardiogram (ECG) interpretation is crucial for identifying these anomalies.
• Blood rheology and flow dynamics are intrinsically linked to cardiac arrhythmias.

Purpose of the Study:

• To estimate cardiac output (CO) using blood flow rate analysis for identifying subjects with LVA.
• To develop and validate a novel algorithm for optimal feature selection in LVA detection.
• To utilize ECG signals for accurate classification of LVA.

Main Methods:

• Cardiac output (CO) estimation derived from stroke volume (SV), end-diastolic/systolic volumes (EDV/ESV), and heart rate derived from ECG R-R intervals.
• Development of the Feature Ranking Score (FRS) algorithm to score and select optimal features from ECG signals.
• Classification of LVA using the Least Square-Support Vector Machine (LS-SVM) classifier with selected features.
• Validation using signals from the public domain MIT-BIH arrhythmia database.

Main Results:

• The study successfully estimated CO as a vital parameter for LVA identification.
• The FRS algorithm effectively identified and selected optimal features for classification.
• The LS-SVM classifier demonstrated proficiency in identifying LVA using the selected features.
• The proposed technique showed validation in identifying LVA from ECG signals and blood flow characteristics.

Conclusions:

• Deviation in CO values from nominal ranges indicates a higher susceptibility to LVA.
• The FRS algorithm combined with LS-SVM provides an effective method for LVA detection.
• This approach offers a promising tool for early identification and management of LVA using ECG and blood flow analysis.