A0674
Title: The mediating role of neuroimaging data in age-related cognitive decline
Authors: Shuo Chen - University of Maryland (United States) [presenting]
Abstract: Understanding age-related cognitive decline has been central to aging neuroscience. Aging affects brain functions and structures and consequently causes decayed neurocognitive performance. To further understand this process, we investigate the mediation role of multivariate neuroimaging variables in age-related cognitive decline. Considering the fact that cognition is exclusively determined by the brain, we propose a new multivariate mediation model by maximizing the mediation effect of brain imaging data. Specifically, we decompose the total effect of aging on cognitive function into natural direct and indirect effects and maximize the indirect effect with a parsimonious set of neuroimaging variables. We implement the optimization problem by alternating the direction method of multipliers and considering the aggregate effect of selected imaging variables as a functional brain age score (FBAS). The simulation results show that our method can accurately select imaging variables and estimate the mediation effect in comparison to existing methods. We further apply the proposed method to the whole-brain cortical thickness and white-mater integrity data of 37,441 UK Biobank participants and found that the mediation effect of brain imaging variables explains more than 90\% of the age effect on cognitive decline.