Linear and non-linear brain-heart and brain-brain interactions during sleep

Area of Science:

• Neuroscience
• Physiology
• Complexity Science

Background:

• Understanding the interplay between cardiac and brain activity during sleep is crucial for comprehending physiological regulation.
• Heart rate variability (HRV) and electroencephalographic (EEG) rhythms offer insights into autonomic and central nervous system functions, respectively.

Purpose of the Study:

• To investigate the physiological networks governing the joint modulation of HRV and EEG rhythms during sleep.
• To compare the efficacy of Granger causality (GC) and transfer entropy (TE) in characterizing these complex interactions.

Main Methods:

• Acquired time series of normalized high-frequency HRV and EEG power spectra (δ, θ, α, σ, β bands) from 10 healthy subjects during sleep.
• Applied GC (linear model-based) and TE (nonlinear model-free) to assess directed interactions between physiological signals, conditional on other series.
• Utilized surrogate data to evaluate the contribution of nonlinear dynamics to TE.

Main Results:

• Both GC and TE identified structured physiological interaction networks.
• Predominant information flow was observed from HRV to EEG rhythms (brain-heart network).
• Within the brain, σ and β EEG waves influenced δ, θ, and α waves, with significant nonlinear contributions from the δ node.

Conclusions:

• GC and TE are suitable for mapping physiological networks, with a notable linear component.
• Nonlinear dynamics, particularly involving the δ EEG band, play a significant role in autonomic and brain regulation during sleep.
• Nonparametric TE estimation is vital for uncovering the intricate structure of these physiological networks.