Shockable Rhythm Diagnosis for Automated External Defibrillators Using a Modified Variational Mode Decomposition Technique

Sudden cardiac arrests are mainly caused by ventricular fibrillation and ventricular tachycardia, which are known as shockable (SH) rhythms and properly curable by electronic defibrillators. In this paper, we propose a novel algorithm to decide whether an electrocardiogram (ECG) signal is SH or nonshockable (NSH). The algorithm selects 20 features from both the preprocessed ECG and its NSH signal using a modified variational mode decomposition technique, and uses the support vector machine for the binary SH/NSH classification. The 20 features are identified as an efficient set of the most informative, among 54 candidate features, by comparing the balanced error rate of each combination, based on two layers of feature selection. This feature set is validated with the evaluation data in the public database using a fivefolds cross validation procedure. The proposed algorithm results in accuracy of 99.00%, sensitivity of 97.36%, and specificity of 99.16%.