Mean field model of alzheimer’s progression biomarkers in EEG

This thesis is intended to show some of the peculiar behavioral changes of a simulated whole brain network after the insertion and/or modification of structural parameters in or- der to describe the insurgency of a prodromal stage of Alzheimer’s disease (AD). AD, even in its preclinical stages, is a complex disease capable of altering both the structure (e.g. the functional connectivity) and the temporal dynamics of the brain network. Nonetheless, the mechanism underlying such alterations are not know as of today, nor a precise whole brain model has ever attempted to capture the faceted anomalies (often considered as AD biomark- ers) induced by such disease in brain signals. The simulations herein presented aim to grasp information (and maybe give some hint of hierarchy) about the role, in prodromal stages of Alzheimer (pAD), of alterations of both structure and dynamics, to the meso (single brain region) and global (whole cortex) scale. Results have been analyzed by comparing the time series of electrophysiological signals between the (both simulated) AD and the healthy case (HC) during a resting state eyes closed EEG exam. What was found, is a complex pattern of impairment in brain activity derived both from the disease’s direct phenomenology (i.e. decreased inhibition and connectivity in regions affected by the disease) and homeostatic adjustments the brain adopts in order to preserve its state (consisting in a global hyperex- citability and in a connection reweighting between brain parts). The model was also able to determine causal relations between the pathophysiological aspects of the disease and evoked EEG biomarkers. Furthermore, a graph-theoretical quantity has been proposed, able to characterize the evolution of the disease in its prodromal phases, by bridging the observed anomalies in simulated brain signals to the underlying structural alterations. This quantity also allowed for a subject-specific model of disease progression, that has been tested by com- parison with EEG recordings from patients previously categorized in many pAD categories. The comparison with such human data corroborates the precision of the model, capable of capturing both the pAD-related EEG biomarkers and the evolutionary path of the disease observed in experimental signals