Detection of sudden cardiac death by a comparative study of heart rate variability in normal and abnormal heart conditions

Background and objective: Sudden cardiac death (SCD) is an unexpected loss in functioning of the heart. It generally occurs within 1 h of onset of symptoms. No reliable method for early detection of SCD is available. Therefore, the development of a non-invasive method for risk identification remains a topic of utmost interest. This paper presents a novel approach for detection of the risk of SCD by performing a comparative analysis of heart rate variability (HRV) in normal subjects as well as patients with coronary artery disease and heart failure. Methods: HRV of four subject groups, normal sinus rhythm, coronary artery disease, congestive heart failure, and SCD, has been analyzed. The analysis was performed by using nonlinear techniques and time-frequency representation obtained by generalized S-transform. The extracted features were examined for their clinical significance by using the Kruskal-Wallis one-way analysis of variance and multiple comparisons. Eventually, classification was performed using support vector machines and decision tree classifiers to separate the individuals at risk of SCD. Results: The performance of the proposed methodology has been evaluated using PhysioNet open-access databases. Statistical analysis shows that HRV in SCD group differs significantly from other groups. For classification, an accuracy of 91.67% was achieved with 83.33% sensitivity, 94.64% specificity, and 84.75% precision. Conclusion: The experimental results obtained by analyzing retrospective data seem promising. However, the methodology needs to be tested on larger databases to generalize the findings. Prospective studies on its clinical usefulness may help in developing a concrete diagnostic technique. (C) 2020 Nalecz Institute of Biocybernetics and Biomedical Engineering of the Polish Academy of Sciences. Published by Elsevier B.V. All rights reserved.