A multivariate and multistate analysis of dynamics of cardiovascular signals

In this paper, in order to understand regulatory mechanisms of the cardiovascular system from an integrative point of view based on its state-dependent dynamics, mutual relationships among multivariate dynamics (inter-R wave interval: RR; systolic blood pressure: BP; stroke volume: SV; respiration rhythm: RP) in slow-wave sleep (SWS), rapid-eye-movement sleep (REM), supine posture (SUP), and standing posture (STD) are investigated. Kullback–Leibler divergence (KLD) is used to quantify the mutual relationship between the state-specific dynamics. The relationship between the multivariate dynamics is represented by KLDs between the dynamics of the same signals in the different states and those between the dynamics of the different signals in the same state. The relationship among the dynamics of the same signals is summarized as follows, which is quantified with reference to the dynamics in SWS. For BP dynamics, SUP and REM are located near SWS. STD is most distant from SWS in the dynamics of BP and RR. The state dependency of SV dynamics is relatively small, compared with the other signals. Concerning the relationship between the dynamics of the different signals, BP tends to be distant from the other signals in SUP and REM, compared with SWS. The distances between SV and the other signals tend to be larger in STD than the other combinations of signals. These findings suggest that the balance among parasympathetic, sympathetic, and hormonal regulation is specific to the physiological state, and the interrelationship among these state-specific balances persists through the subjects. In addition, trajectories of relationship among dynamics of cardiovascular signals are obtained during state transitions. The trajectories commonly show that the relationship is drastically altered around the boundary between the physiological states. This study is expected to provide a novel framework for exploring regulatory mechanisms of the cardiovascular system and extracting clinical information based on the dynamics analysis.