Use of Spatial- and Connectivity-based Cortical Brain Region Information in Regularized Regression

Recently developed regularization methods utilize information about the brain structure to improve the coefficient estimation in regression models. Our method incorporates structural connectivity and cortical distance information in the penalty term via a Laplacian matrix. The performance of the proposed approach in evaluated via extensive simulation studies and an application to the data from the Human Connectome Project.

Speaker: Jaroslaw Harezlak, PhD, Professor of Epidemiology & Biostatistics, School of Public Health - Bloomington, University of Indiana

Register: http://tiny.cc/EpiBioEmailList