MULTI-SCALE DYNAMICS OF SPONTANOUS BRAIN ACTIVITY CORRELATE WITH WALKING SPEED IN COMMUNITY-DWELLING OLDER ADULTS

摘要

The control of walking depends upon the exchange of information across numerous brain networks. Within a given brain region, spontaneous (i.e., resting-state) fluctuations in neuronal activity are “complex,” containing meaningful information over multiple temporal scales. While this complexity appears to decline with advancing age, it is unknown if it associates with walking performance. The aim of this study was to determine the relationship between resting-state complexity and walking speed in older adults. Forty-three older adults from the MOBILIZE Boston Study completed a resting-state MRI and on a separate visit, assessment of walking under normal conditions (i.e., single-task) and while performing a cognitive serial-subtraction task (i.e., dual-task). Resting-state complexity within 50 pre-defined brain regions was calculated by computing the ‘multiscale entropy’ of the blood-oxygen-level-dependent time-series of each voxel, and then averaging across all voxels of each region. A machine learning technique termed ‘leave-one-subject-out support vector regression,’ adjusted for age, indicated that the estimation of both single- and dual-task walking speed based upon regional resting-state complexity was high (r>0.38, p<0.007). The complexity of seven clusters with known involvement in motor control, attention and/or executive function correlated with both single- and dual-task walking speed (mean p<0.05, mean r>0.31). Four additional regions associated with attentional control were correlated with dual-task walking speed only (mean p<0.03, mean r>0.35). This study revealed that in older adults, walking speed correlates with region-specific multi-scale dynamics of resting-state brain activity, and dual task walking may be particularly dependent upon brain regions that give rise to the control of attention