Network Spreading Dynamics in Cognition and Neurodegeneration

The macroscale network structure of the brain is fundamental to the pathophysiology and treatment of several neuropsychiatric diseases, including epilepsy, neurodegenerative disease, depression, and psychosis. Functional interactions at this scale index disease symptoms and guide exogenous interventions, such as brain stimulation or pharmacology. However, a lack of tools for measuring the underlying neurobiological drivers of these functional interactions, as well as phenotypic heterogeneity within disorders, hinders the ability to expand upon existing treatments and target them to the appropriate populations. Dynamical systems models have the potential to move beyond a statistical description of neural systems by positing mechanisms that link the physical form of a system with its emergent function. A subset of dynamical systems models, linear network spreading models, have proved especially useful for capturing activity fluctuations in neural systems. Existing tools allow these models to be studied through the lens of network control theory, which captures the system’s response to external inputs. Here, we use network spreading models and other computational tools to study structure-function relationships in the human brain and the mechanisms of Parkinson’s disease pathophysiology. First, we employed a network spreading model to characterize the neural substrates of individual differences in impulse control throughout development. Second, we incorporated external inputs into those models in order to explain the temporal progression of large- scale cortical activity patterns. Third, we used a network spreading model to confirm that endogenous levels of α-synuclein, along with both anterograde and retrograde transsynaptic diffusion drive Parkinson’s disease progression. Finally, we identify latent patterns of co- occurring pathologies in neuropathological autopsy data that can be predicted from in vivo biomarkers using statistical models. This collection of studies expands our understanding of how brain activity and misfolded proteins spread throughout the brain’s white matter connections and provides a computational framework for addressing heterogeneity in neurologic diseases. These findings are complementary to the aim of developing network-oriented therapies and lay a general framework for parsing disease heterogeneity across multiple fields of medicine