Title: Estimating heterogeneous biomarker networks and effects on disease outcomes
Authors: Yuanjia Wang - Columbia University (United States) [presenting]
Abstract: Biomarkers are often organized into networks, in which the strengths of network connections vary across subjects depending on subject-specific covariates (e.g., genetic variants). Variation of brain network connections, as subject-specific feature variables, has been found to predict disease clinical outcomes. We develop a two-stage statistical method to estimate covariate-dependent brain networks to account for heterogeneity among network measures and evaluate their association with disease clinical manifestation. In the first stage, we propose a conditional Gaussian graphical model with mean and precision matrix depending on covariates to obtain subject- specific networks. In the second stage, we associate a subject biomarker network measure (connection strengths) estimated from the first step along with the biomarkers and covariates to identify important features of a clinical outcome. The second stage allows us to evaluate the improvement in predictiveness of adding network measures compared to using biomarkers and covariates alone. We assess the performance of the proposed method by extensive simulation studies and apply the method to a Huntington's disease (HD) study to investigate the effect of the HD causal gene on the rate of change in motor symptom as mediated through brain subcortical and cortical gray matter atrophy connections.