Using Wavelet Transform and Machine Learning to Predict Heart Fibrillation Disease on ECG

Abstract

Processing ECG signals with high-frequency low-amplitude sections is a laborious task. Since this process is usually performed by specialists - doctors visually, the possibility of obtaining an incorrect interpretation of the ECG image is not ruled out. ECG is a method of studying the bioelectric activity of the heart. The research method is based on the graphical registration of the received bioelectric signals. In this connection, it became necessary to search for new methods for predicting signal propagation in various directions of science. The problems of extracting information from the electrophysiological signal that can not be obtained by visual analysis of the record, as well as the problems of automation of traditional algorithms of medical analysis are relevant in connection with the lack of research in this field. In this paper, we consider the approach of automatic electrocardiographic signals interpretation of cardiac valves based on the wavelet transform method. The model of the neural network of wavelet packets developed by us is used. The productivity of the constructed system was evaluated on more than 8000 samples. Test results showed that this system was effective when using wavelet transformation. The correct rate of classification was about 95.6 percent.