Understanding Schizophrenia Pathophysiology via fMRI-Based Information Theory and Multiplex Network Analysis.

This work investigates the mechanisms of information transfer underlying causal relationships between brain regions during resting-state conditions in patients with schizophrenia (SCZ). A large fMRI dataset including healthy controls and SCZ patients was analyzed to estimate directed information flow using local Transfer Entropy (TE). Four functional interaction patterns-referred to as rules-were identified between brain regions: activation in the same state (ActS), activation in the opposite state (ActO), turn-off in the same state (TfS), and turn-off in the opposite state (TfO), indicating a dynamics toward converging (Acts/Tfs = S) and diverging (ActO/TfO = O) states of brain regions. These interactions were integrated within a multiplex network framework, in which each rule was represented as a directed network layer. Our results reveal widespread alterations in the functional architecture of SCZ brain networks, particularly affecting schizophrenia-related systems such as bottom-up sensory pathways and associative cortical dynamics. An imbalance between S and O rules was observed, leading to reduced network stability. This shift results in a more randomized functional network organization. These findings provide a mechanistic link between excitation/inhibition (E/I) imbalance and mesoscopic network dysconnectivity, in agreement with previous dynamic functional connectivity and Dynamic Causal Modeling (DCM) studies. Overall, our approach offers an integrated framework for characterizing directed brain communication patterns and psychiatric phenotypes. Future work will focus on systematic comparisons with DCM and other functional connectivity methods.