Electrocardiogram and Phonocardiogram Signal Data Fusion Using Deep Learning System

In this paper, we will analyze a relevant topic in the field of biomedicine: the fusion of information when different types of analyses or signals are used to classify a pathology. In our case, we will perform an early fusion system using the electrocardiogram and the phonocardiogram signal. It is logical that there is a certain correlation between both signals, as professionals have used both to determine and predict pathologies. In this article, we propose to merge both sources of information to develop an automated classification aid system that enables collaboration with medical experts using deep learning systems. The first step in the presented methodology is to treat the two signals independently. For this purpose, we will use the information from the scalogram based on the discrete wavelet transform for an Electrocardiogram (ECG) and for an Phonocardiogram (PCG). Once this information is available, it will be compared between isolated deep learning systems and a combined system. The advantages and superior performance of the combined information system are analyzed in this article.