Designing a Convolutional Neural Network for Classifying Phonocardiogram Signals for Coronary Artery Disease Detection

In this paper we propose an automatic and efficient method for diagnosing heart diseases, aimed at reducing errors by utilizing artificial neural networks. The primary goal is to minimize risks and save more lives through accurate and early diagnoses. Consequently, this approach reduces the time and costs associated with manual analysis while enabling accurate and rapid assessments. This method is designed for classifying phonocardiogram (PCG) signals. In this method, a onedimensional convolutional neural network (1D-CNN) is trained to analyze PCG signals by comparing them to those from healthy individuals and patients, helping the detection of coronary artery disease (CAD). Although several automatic cardiovascular disease (CVD) screening methods have been proposed, this study introduces an efficient process that provides better accuracy in distinguishing healthy hearts from those affected by CAD. Comparative analysis with existing methods shows that our proposed deep learning model for heart sound classification provides better performance, as reflected in enhanced classifier metrics.