Standard multiscale entropy reflects neural dynamics at mismatched temporal scales: What’s signal irregularity got to do with it?

• DOI: 10.1371/journal.pcbi.1007885
• Publication Date: 2020-05-11T18:09:24Z
• Authors: Julian Q. Kosciessa; Niels A. Kloosterman; Douglas D. Garrett
• Keywords: Surrogate data; Sample entropy

Multiscale Entropy (MSE) is used to characterize the temporal irregularity of neural time series patterns. Due its’ presumed sensitivity non-linear signal characteristics, MSE typically considered a complementary measure brain dynamics variance and spectral power. However, divergence between these measures often unclear in application. Furthermore, it commonly assumed (yet sparingly verified) that entropy estimated at specific scales reflects those precise function. We argue such assumptions are not tenable. Using simulated empirical electroencephalogram (EEG) data from 47 younger 52 older adults, we indicate strong previously underappreciated associations power, highlight how links preclude traditional interpretations scales. Specifically, show typical definition patterns via “similarity bounds” biases coarse scales–that thought reflect slow dynamics–by high-frequency dynamics. Moreover, demonstrate fine scales–presumed fast dynamics–is highly sensitive broadband dominated by low-frequency contributions. Jointly, issues produce counterintuitive reflections frequency-specific content on emphasize resulting inferential problems conceptual replication cross-sectional age differences rest, which scale-specific effects could be explained power mismatched may alleviated, indication rhythmic irregularity. By controlling for narrowband contributions, spontaneous alpha rhythms during eyes open rest transiently reduce Finally, recommend best practices better permit valid estimation interpretation interest.