Time-Resolved Brain-to-Heart Probabilistic Information Transfer Estimation Using Inhomogeneous Point-Process Models

Objective:
The quantification of functional brain–heart interplay through the dynamics of the central and autonomic nervous systems may provide effective biomarkers for cognitive, emotional, and autonomic state changes. Despite several computational models were proposed to this end, none provides a directional estimation of such interplay in a time-resolved and probabilistic fashion.
Methods:
In this study, a multivariate inhomogeneous point-process model for heartbeat dynamics is employed to derive subject-specific, time-resolved, functional estimates of the directional interplay occurring from the brain to the heart, whose activity is represented by electroencephalography and R-peaks intervals series. An inverse-Gaussian probability density function is used to predict heartbeat events as a function of neural dynamics, which is modeled as an exogenous input to the autoregressive cardiac dynamics.
Results:
The performance is evaluated using heart rate variability and electroencephalography series gathered from 24 healthy volunteers undergoing a cold-pressor test, and the modeling goodness-of-fit is assessed through the time-rescaling theorem. The results suggest that cortical dynamics drives heartbeat series with specific time delays in the range of 30s to 60s and 90s to 120s from the peripheral thermal stress onset.
Conclusion:
The proposed framework provides novel insights in human neurophysiology, exploiting a fully probabilistic definition of the continuous functional brain–heart interplay.