Deep Learning Framework with ECG Feature-Based Kernels for Heart Disease Classification

Abstract

Heart disease classification with high accuracy can support the physician’s correct decision on patients. This paper proposes a kernel size calculation based on P, Q, R, and S waves of one heartbeat to enhance classification accuracy in a deep learning framework. In addition, Electrocardiogram (ECG) signals were filtered using wavelet transform with dmey wavelet, in which the shape of the dmey is closed to that of one heartbeat. With this selected dmey, each heartbeat was standardized with 300 samples for calculation of kernel sizes so that it contains most features in each heartbeat. Therefore, in this research, with 103,459 heart rhythms from the MIT-BIH Arrhythmia Database, the proposed approach for calculation of kernel sizes is effective with seven convolutional layers and other fully connected layers in a Deep Neural Network (DNN). In particular, with five types of heart disease, the result of the high classification accuracy is about 99.4 %. It means that the proposed kernel size calculation in the convolutional layers can achieve good classification performance and it may be developed for classifying different types of disease.

Wavelet Scattering Transform for ECG Cardiovascular Disease Classification

“…Future work suggestions include employing data augmentation techniques and exploring deep-learning optimization methods for enhanced ARR classification. According to Nguyen et al [15]. presented a deep learning framework to improve CVD classification accuracy.…”

A Novel Seismocardiogram Mathematical Model for Simplified Adjustment of Adaptive Filter

“…𝑑 𝑛 𝑦 𝑛 (12) where: e(n)-adaptation error, y(n)-denoised signal in output, d(n)-desired signal. The operation of this filter is based on the tendency of the filter output signal y(n) defined by Equation ( 13) to correspond as closely as possible to the affected signal d(n) through the feedback response of the error signal e(n) to the coefficients H(z) of the filter transfer function, so that e(n) is zero [43,44].…”

“…These portable or wearable electronic systems do not disturb the daily activity of patients, have evaluated data noninvasively and extract various information on heart functionality [9,10]. The seismocardiogram signal represents low-frequency (0-50 Hz) cardio cycle mechanic vibrations corresponding to cardiac events [11,12], and contains mitral valve opening (MO) and closing (MC), isovolumetric contraction, ejection, opening of the aortic valve (AO) and closing (AC), and cardiac filling [13]. The heart monitoring system is complicated due to its nonstationary nature [14] and the presence of noises such as muscle movement noise, respiration vibration noise [8], and environment acoustic noise [6,15] in the seismocardiogram (SCG) signal [16,17].…”

Driver Cardiovascular Disease Detection Using Seismocardiogram

“…This transient loss of consciousness event is called syncope and can result from cardiac conditions, mostly consisting of arrhythmic events, bradyarrhythmia, or tachyarrhythmias [8,9]. For this reason, the American Heart Association and Heart Rhythm Society developed recommendations to prevent a person from putting others at risk of harm and at the same time have the opportunity to work according to a society and its culture [1,7,10,11].…”