|
Abstract:
|
Various brain recordings and biological techniques have been developed to systematically examine how genetic factors and brain function jointly affect human behaviors and cognitive functions. These types of multi-modal data are large, high dimensional, endowed with a complicated dependence structure and low signal-to-noise ratio. To efficiently model the genetic effects on brain signal data, we developed a novel random-effect tensor regression with separable covariance. We derived and compared several estimating algorithms to estimate genetic parameters of interest, such as genetic heritability. Furthermore, we investigated the genetic contribution to brain connectivity, which characterizes the dependence between different brain regions, during a working memory study.
|
