A0470
Title: Partition learning for functional neuro connectivity
Authors: Emily Hector - North Carolina State University (United States) [presenting]
Abstract: Motivated by the need to model joint dependence between regions of interest in functional neuro connectivity for efficient inference, a new Bayesian clustering approach is proposed for correlation structures of high-dimensional Gaussian outcomes. The key technique is a Dirichlet process that clusters correlation sub-matrices into independent subgroups of outcomes, thereby naturally inducing sparsity in the whole brain connectivity matrix. A new split-merge algorithm is employed to improve the mixing of the sampling chain shown empirically to recover both uniform and true Dirichlet partitions with high accuracy. The approach's performance is investigated through extensive simulations. Finally, the proposed approach is used to group regions of interest into functionally independent sub-groups in the Autism Brain Imaging Data Exchange participants with autism spectrum disorder and attention-deficit/hyperactivity disorder.
