A0751
Title: Network on network regression
Authors: Shuo Chen - University of Maryland (United States) [presenting]
Abstract: A new statistical model is presented for multivariate edge outcomes in a network and multiple edge predictors in a network. The research is motivated by the association analysis between brain network data. For example, the interest is to test how resting-state functional connectome influences the task-related functional connectome and how structural connectivity network affects functional connectivity network. In these analyses, nodes denote brain regions, and edges represent connections. A triple-layer network model is proposed to capture the complex associations and identify dense association subgraph pairs. The proportion of related edges is maximized in both outcome and predictor subgraphs while ensuring that most related edges are covered by the subgraph pairs. Extensive simulations are conducted, and the proposed method is applied to data from the human connectome project and UK Biobank.
