Modeling Alzheimer's Disease: Bayesian Copula Graphical Model from Demographic, Cognitive, and Neuroimaging Data

The early detection of Alzheimer's disease (AD) requires the understanding of the relations between a wide range of disease-related features. Analyses that estimate these relations and evaluate their uncertainty are still rare. We address this gap by presenting a Bayesian approach using a Gaussian copula graphical model (GCGM). This model is able to estimate the relations between both continuous, discrete, and binary variables and compute the uncertainty of these estimates. Our method estimates the relations between brain-region specific gray matter volume and glucose uptake, amyloid levels, demographic information, and cognitive test scores. We applied our model to 1022 participants across different stages of AD. We found three indirect pathways through which old age reduces cognition: hippocampal volume loss, posterior cingulate cortex (PCC) volume loss, and amyloid accumulation. Corrected for other variables, we found that women perform better on cognitive tests, but also discovered four indirect pathways that dampen this association in women: lower hippocampal volume, lower PCC volume, more amyloid accumulation and less education. We found limited relations between brain-region specific glucose uptake and cognition, but did discover that the hippocampus and PCC volumes are related to cognition. These results showcase that the novel use of GCGMs can offer valuable insights into AD pathogenesis.