Decomposing the transfer entropy to quantify lag-specific Granger causality in cardiovascular variability

Area of Science:

• Information theory
• Complex systems analysis
• Time series analysis

Background:

• Transfer entropy (TE) quantifies information flow between coupled processes.
• Existing TE methods lack precise temporal resolution for information transfer.
• Understanding directed information flow with exact timing is crucial for complex systems.

Purpose of the Study:

• To develop a modified transfer entropy (TE) capable of detecting the exact timing of information transfer.
• To propose a method for estimating lag-specific TE and relevant time delays from time series data.
• To validate the proposed approach in simulations and real-world cardiovascular data.

Main Methods:

• Decomposition strategy based on lag-specific Granger causality to identify significant time delays.
• Assigning information transfer amounts to specific delays to reconstruct overall TE.
• Procedures for practical estimation of lag-specific TE in bivariate and multivariate settings.

Main Results:

• Demonstrated feasibility of lag-specific TE estimation from time series data.
• Successfully reflected known cardiovascular regulatory mechanisms in real data.
• Highlighted the necessity of multivariate TE for accurate time-lagged information transfer assessment in multi-system interactions.

Conclusions:

• The proposed lag-specific TE method accurately quantifies the timing and direction of information transfer.
• This approach provides a powerful tool for analyzing complex systems, particularly in physiological regulation.
• Multivariate lag-specific TE is essential for comprehensive understanding of information flow in interconnected systems.