Abstract
<jats:p>Background: Balance maintenance in humans is not only a mechanical process, but it also relies on continuous interactions between cortical activity and body dynamics. Alterations in postural sway are commonly observed in aging and stroke and are frequently used to assess balance impairment. However, similar balance deficits do not necessarily reflect similar underlying sensorimotor control mechanisms. Therefore, investigating brain body interactions and their relationship to characteristics of postural behavior may provide deeper insights into the neural processes underlying balance dysfunction in these populations. Objective: To determine whether brain body coupling is associated with characteristics of postural behavior captured by the temporal organization of postural fluctuations beyond conventional magnitude based measures of postural sway, and whether these relationships differ between stroke survivors, healthy older adults, and young adults. Methods: EEG and center of pressure (CoP) signals were recorded simultaneously in stroke survivors (n = 12), healthy older adults (n = 18), and young controls (n = 17) during quiet standing under 4 different manipulated sensory conditions. Sway based corticokinematic coherence (CKC) as well as linear and nonlinear features (sample entropy, SE; fractal dimension, FD) of CoP were extracted. Linear mixed-effects model assessed associations between features and CKC, and model performance was compared using Akaike Information Criterion. Multidimensional state vectors were constructed from CKC, linear and nonlinear CoP features, and Euclidean distances between consecutive states in the standardized feature space were computed to quantify condition dependent transitions in brain body control organization. Results: Nonlinear features showed significant, group and feature dependent associations with CKC in the mediolateral direction, driven by significant SE and FD effects in the stroke group and an SE effect in the older group, while no significant associations were observed in the young group. Including nonlinear features in baseline models containing only linear CoP features significantly improved model fit. CKC alone showed low classification performance (AUC 50 to 65), whereas combining CKC with linear and nonlinear features improved group discrimination (AUC up to 0.86). State space transition analysis revealed larger condition dependent transitions in stroke participants compared with healthy older adults, particularly going from eyes open to eyes closed when standing on foam. Conclusion: Brain body coupling during standing may be understood more comprehensively by factoring in the temporal structure of fluctuations rather than their amplitude alone. These findings support the use of nonlinear dynamical features, combined with CKC, as potential markers of balance impairment.</jats:p>