A Fast and Efficient Ensemble Transfer Entropy and Applications in Neural Signals

Area of Science:

• Neuroscience
• Computational Neuroscience
• Information Theory

Background:

• Ensemble transfer entropy (TEensemble) analyzes dynamic interactions between brain regions.
• Traditional TEensemble is computationally intensive due to extensive surrogate data requirements.

Purpose of the Study:

• To develop a computationally efficient TEensemble method.
• To reduce the complexity of estimating transfer entropy from multiple realizations.
• To validate the novel method's performance against traditional approaches.

Main Methods:

• Proposed a fast, efficient TEensemble utilizing a single set of surrogate data for null hypothesis testing.
• Compared the novel TEensemble with the traditional method using simulated neural signals.
• Validated the method's effectiveness on actual neural signals.

Main Results:

• The novel TEensemble reduces computation time by two to three orders of magnitude.
• Accurately tracks dynamic interactions, detecting interaction strength and direction robustly, even with noise.
• Achieves steady state slower than the traditional method with increased samples.

Conclusions:

• The proposed efficient TEensemble offers a computationally feasible approach for investigating dynamic brain region interactions.
• This method enhances the practical application of transfer entropy in neuroscience.
• Provides a robust tool for analyzing complex neural communication patterns.