Hybrid Learning Approach for Automated Identification and Categorization of Cardiovascular Disorders

Abstract

ardio Vascular Diseases (CVDs) pose an important global health challenge, contributing substantially to mortality rates worldwide. Electrocardiography (ECG) is a necessary diagnostic tool in the detection of CVDs. Manual analysis by medical experts, for ECG interpretation, is laborious and subject to interobserver variability. To overcome these limitations, automated categorization technique has gained prominence, enabling efficient CVDs classification. The major focus of this work is to utilize deep learning (DL) approach for the identification of CVDs using ECG signals. The presented work incorporates two hybrid models: one-dimensional convolutional neural network (1D-CNN) with Recurrent Hopfield Neural Network (1DCNN-RHNN) and with Residual Network (1D-CNN-ResNet), to obtain important features from raw data and categorize them into different groups that correlate to CVD situation. The 1D-CNN-RHNN model achieved classification accuracy of 96.62% in the 4-class classification of normal, coronary artery disease (CAD), myocardial infarction (MI), and congestive heart failure (CHF) and the 1DCNN-ResNet model achieved classification accuracy of 95.75% in the 5-class classification of normal, CAD, MI, CHF and cardiomyopathy. The proposed model's functionality is validated with medical ECG data, and its outcomes are evaluated using various measures. Experimental findings demonstrate that the proposed models outperform other existing approaches in categorizing multiple classes. Our suggested approach might potentially help doctors screen for CVDs using ECG signals and is capable of being verified with larger databases.