Comparison of Linear Model-Based and Nonlinear Model-Free Directional Coupling Measures: Analysis of Cardiovascular and Cardiorespiratory Interactions at Rest and During Physiological Stress

Abstract

In this work, we present an investigation of the cardiovascular and cardiorespiratory regulatory mechanisms involved during stress responses using the information-theoretic measure of transfer entropy (TE). Specifically, the aim of the study is to compare different estimation approaches for the evaluation of the information transferred among different physiological systems. The analysis was carried out on the series of heart period, systolic arterial pressure and respiration measured from 61 young healthy subjects, at rest and during orthostatic and mental stress states, by using both a linear model-based and a nonlinear modelfree approaches to compute TE. The results reveal mostly significant correlations for the measures of TE estimated with the two approaches, particularly when assessing the influence of respiration on cardiovascular activity during mental stress and the influence of vascular dynamics on cardiac activity during tilt. Therefore, our findings suggest that the simpler linear parametric approach is suitable in conditions predominantly regulated by sympathetic nervous system or by the withdrawal of the parasympathetic system.