Time-Resolved Directional Brain-Heart Interplay Measurement Through Synthetic Data Generation Models

Although a plethora of synthetic data generation models have been proposed to validate biomarkers of brain and cardiovascular dynamics separately, a limited number of computational methods estimating directed brain-heart information flow are currently available in the scientific literature. This study introduces a computational framework exploiting existing generative models for a novel time-resolved quantification of causal brain-heart interplay. Exemplarily, having electroencephalographic signals and heart rate variability series as inputs, respective synthetic data models are coupled through parametrised functions defined in accordance with current central autonomic network (CAN) knowledge. We validate this concept using data from 30 healthy volunteers undergoing notable sympathetic elicitation through a cold-pressor test, and further compare the obtained results with a state-of-the-art method as maximal information coefficient. Although our findings are in agreement with previous CAN findings, we report new insights into the role of fronto-parietal region activity and lateralisation mechanisms over the temporal cortices during prolonged peripheral elicitation, which occur with specific time delays. Additionally, the afferent autonomic outflow maps to brain oscillations in the δ and γ bands, whereas complementary cortical dynamics in the θ, α, and β bands act on efferent autonomic control. The proposed framework paves the way towards novel biomarker definitions for the assessment of complex physiological networks using existing data generation models for brain and peripheral dynamics.