Exploring the Reconfiguration Effects of Physical Fatigue on the Gradient Space of Brain Function Based on EEG

This study examined how physical fatigue reorganizes the low-dimensional structure of EEG functional connectivity using three embedding approaches.

Thirty male participants completed a Bruce treadmill test to exhaustion, with resting-state EEG recorded before and after exercise. Frequency-specific wPLI connectivity matrices were analyzed using PCA, Laplacian eigenmaps, and diffusion maps with common-reference alignment. Regional G1-G3 scores, G1-G2 displacement, λ1-λ3 eigenvalues, and their associations with exercise duration were assessed.

PCA showed increased beta-band G1 scores over the central scalp region and decreased scores over the occipital region, together with reduced alpha-band G3 over the central region. LE showed increased theta-band G3 over the central region. PCA displacement exhibited a significant frequency band × scalp region interaction, driven by greater alpha-band occipital displacement. Theta-band λ1-λ3 increased with PCA but decreased with DM, whereas LE showed no significant eigenvalue changes. The PCA-derived beta-band central ΔG1 was positively correlated with exhaustive exercise duration (r = 0.50, p = 0.004).

Physical fatigue induces frequency-, region-, component-, and method-specific reorganization of the low-dimensional structure of scalp-level EEG functional connectivity. PCA was particularly sensitive to spatially differentiated gradient changes, while LE and DM captured complementary aspects of fatigue-related reorganization. The association between central beta-band ΔG1 and exercise duration further suggests potential behavioral relevance of gradient.