B0847
Title: A simple approach to modeling pre- vs post-treatment differences in functional connectivity
Authors: Hyung Park - New York University School of Medicine (United States) [presenting]
Abstract: The challenge of modelling treatment effects is addressed on complex data objects, such as functional connectivity matrices. The proposed method focuses on analyzing changes in functional connectivity before and after treatment. It achieves this by parametrizing pre- and post-treatment covariances in a common tangent space and performing a data-driven projection of the response signals to capture treatment-related changes in the covariances. The method incorporates a matrix whitening approach to account for individual variations. A Bayesian framework enables joint estimation of parameters and uncertainty quantification, combining repeated measure model estimation with dimension reduction for covariance matrices. The method is applied to analyze treatment associations with functional connectivity in a depression clinical trial dataset.
