An Approach to Decompose the Information Shared between Cardiovascular and Respiratory Physiological Systems

The framework of Partial Information Decomposition (PID) quantifies the redundant and synergistic information that two source variables share with a third variable within a network of interacting systems.While the PID has been formulated so far either for discrete or continuous Gaussian variables, this work introduces a computation strategy, based on nearest-neighbor entropy estimation, to decompose the information that a discrete variable shares with two continuous sources.The approach is first validated in a Boolean logic gate producing exclusive synergy, and then applied in the context of cardiovascular physiology to decompose the information shared by the binary variable mapping the respiratory phase and the continuous time series of heart period and systolic arterial pressure variability.We show how cardiac and vascular regulatory mechanisms are individually related to respiration at rest and during postural stress, and jointly interact in maintaining cardiovascular and cardiorespiratory homeostasis during the inspiratory and expiratory breathing phases.