Nonlinear Pulse Wave Dynamics in Prediction of Coronary Heart Disease and Myocardial Infarction

Abstract

Atherosclerosis may induce coronary heart diseases (CHD) that progress to myocardial infarction (MCI), if untreated. The pulse plethysmogram (PPG) plot demonstrated the change in peripheral blood volume influence induced by the sympathetic branch of the autonomic nervous system that causes vasoconstriction. Thus, the present work has been aimed at proposing a noninvasive, low-cost, and wearable technology for the forecasting of CHD and MCI subjects using PPG-derived features with a support vector machine (SVM). A total of 70 volunteers (50-55 years) including MCI (n = 10), CHD (n = 30), and Control (n = 30) has been participated. Digital PPG was recorded for 10 min from the index finger of the left hand in the supine position. Ten samples from each subject were selected from the PPG signal for deriving nonlinear pulse rate variability (PRV) features using Kubios 2.0. The sensitivity, specificity, and accuracy were calculated from the obtained confusion matrix in the classification of control, CHD, and MCI classes. The findings suggested that the reduced value of PRV features in the Poincare plot, detrended fluctuation analysis recurrence plot, and correlation dimension while higher entropy measures in MCI than the CHD subjects. The best accuracy of 98.2% and 98.3% were observed with quadratic functions in the prediction of CHD and MCI subjects from the control group, respectively. The obtained results suggested the applicability of nonlinear PRV features in designing the low-cost real-time wearable technique in the prediction of CHD and MCI subjects.