Analysis of The Detection of Ventricular Fibrillation in Its First 3 Seconds Using Different Features and Classifiers

This paper presents a method for detecting ventricular fibrillation using features in the time and frequency domains, and measures of signal complexity. Several classifiers and several classification methods are tested, depending on the purpose pursued. The ECG signal segments on which the classification is performed have a duration of 3 seconds. The accuracy for classification into 2 classes (normal sinus rhythm and ventricular fibrillation) depends on the classifier used, and reached 94.6% with the ensemble classifier.