Hierarchical Graphical Model for Learning Functional Network Determinants

Analysis of brain functionality is a stimulating research topic from both a neuroscientific and statistical perspective. Although several works have improved our comprehension of the relationship between subject-specific information and brain architecture, many questions remain open. The aim of this paper is to relate functional connectivity patterns with subject-specific features and brain constraints, such as age and mental illness of the subject and lobes membership for brain regions, and illustrate whether these phenotypes affect the neurophysiological dynamics. To address such goal we consider a modular approach that allows to remove noise from the fMRI data, estimate the functional dependency structure and relate functional architecture with structural and phenotypical information.