Nonparametric Group Variable Selectionwith Multivariate Response forConnectome-Based Prediction of Cognitive Scores

In this article, we study possible relations between the structural connectome and cognitiveprofiles using a multi-response nonparametric regression model under group sparsity. The aim is to identify the brain regions having significant effect on cognitive functioning. In this article, we consider nine different attributesfor each brain region as our predictors. Here, each node thus correspond to a collectionof nodal attributes. Hence, these nodal graph metrics may naturally be grouped togetherfor each node, motivating us to introduce group sparsity for feature selection. We proposeGaussian RBF-nets with a novel group sparsity inducing prior to model the unknown meanfunctions. The covariance structure of the multivariate response is characterized in termsof a linear factor modeling framework. For posterior computation, we develop an efficientMarkov chain Monte Carlo sampling algorithm. We show that the proposed method per-forms overwhelmingly better than all its competitors. Applying our proposed method to aHuman Connectome Project (HCP) dataset, we identify the important brain regions andnodal attributes for cognitive functioning, as well as identify interesting low-dimensionaldependency structures among the cognition related test scores.