Grouped integrative dynamic factor models for intra-and inter- individual brain networks

Abstract: We introduce a novel framework for dynamic factor model-based data integration of multiple subjects, Grouped Integrative Dynamic factor models (GRIDY). This work is a pilot study to suggest empirics where the model can be applied to the multi-block data collected from the pre-labeled and pre-grouped subjects. The study has been designed to focus on figuring out the intra[1]difference between the pre-labeled two groups as a combination of spatial information of each group and individual temporal dependence. Simultaneously, inter-difference within each individual can be explored by investigating different configurations of latent factor series. A novel principal angle-based rank selection algorithm and a well-known non-iterative integrative analysis framework are combined in the methodological aspect. In addition, this framework can reconstruct identifiable latent factor series with flexible covariance structures inspired by the multi-way array decomposition. Simulations under various scenarios and analysis of resting state functional MRI data collected from multiple subjects in both the group in Autism Spectrum Disorder and the control have been conducted.

Share this Event