Ventricular Arrhythmia Classification Using Classical and Neural Network Approaches

The article is dedicated to the classification of ventricular arrhythmias using classical and neural network methods. The study utilized the MIT-BIH Malignant Ventricular Ectopy Database, which contains 22 recordings with various types of arrhythmias. The data were segmented into 2-s fragments and annotated, allowing for the identification of 6 classes of arrhythmias. Various algorithms were applied to solve the classification task, including k-nearest neighbours, support vector machine, and deep neural networks. The results showed that using the smoothed electrocardiogram signals spectrum allows for achieving high sensitivity and classification accuracy. The results of the experiments confirmed the effectiveness of both classical and neural network approaches.