Abstract

We present a neuroimaging-driven study examining the relationship between functional connectivity across cognitive control domains and cognitive phenotypes, aiming to identify specific brain regions significantly associated with these phenotypes. We propose a generalized linear modeling framework incorporating multiple network responses and predictors, allowing for diverse interconnections between edges. Leveraging hierarchical Bayesian modeling, our approach estimates regression coefficients and identifies predictor-linked nodes with precise uncertainty quantification. Empirical investigations, including simulation studies and functional connectivity data analysis, demonstrate our framework's superior performance compared to competitors.