A0561
Title: Covariate-assisted community detection for functional brain networks: A variational approach
Authors: Panpan Zhang - Vanderbilt University Medical Center (United States) [presenting]
Abstract: Understanding the human brain requires modeling its complex and high-dimensional architecture. Graph-based methods for brain networks, particularly community detection techniques, offer critical insights into the modular organization of brain function across different disease stages. A novel covariate-assisted community detection method is proposed that jointly leverages network topology and informative node-level attributes. The method is formulated within a variational estimation framework, enabling efficient estimation in large-scale networks. Through extensive simulation studies, the advantages of the approach are demonstrated over existing alternatives in terms of accuracy and robustness. The proposed method is further applied to fMRI data from individuals with Alzheimer's disease, uncovering meaningful community structures that reflect disease-related alterations in brain connectivity.
