An information decomposition strategy to dissect dynamic short-term cardiovascular and cerebrovascular interactions

This work proposes an approach combining multivariate time series analysis and information dynamics for the joint analysis of short-term cardiovascular (CV) and cerebrovascular (CB) interactions. The approach employs vector autoregressive models to quantify the information provided by mean and systolic arterial pressure on the variability of heart period and cerebral blood flow, conditioning on respiration effects, and applies Partial Information Decomposition (PID) to quantify contributions pertaining separately to CV or CB mechanisms or reflecting the overlap and the interplay between CV and CB mechanisms. The application to variability series measured in patients suffering from postural syncope reveals patterns of the PID measures distinctive of the response to headup tilt and of the transition towards syncope onset.