Methods and algorithms for classifying ECG characteristics and training neural networks for diagnosing heart diseases

Abstract

Purpose to improve methods, develop algorithms for identifying ECG characteristics for automatic diagnosis of heart diseases such as arrhythmia. In this regard, we have proposed methods and algorithms: preprocessing of signals, extraction of functions Q, R, S, P, T; training of neural networks for diagnosing heart diseases; classifying the characteristics of the electrocardiogram for various possible states of the cardiovascular system using neural networks, determining the time and amplitude characteristics of the ECG functions, calculating the intervals between them and the heart rate.The results of the research are algorithms for training neural networks and a method for classifying ECG characteristics for diagnosing heart diseases, implemented on the Matlab programming system.