Computational whole-brain model of progression to dementia accounts for event-related potentials non-linear dynamics during attention task

1 / 1

Attention and memory tasks and associated event-related potentials (ERPs) are gaining popularity as early biomarkers for dementia. The 3-choice vigilance task (3CVT), a test designed to estimate the level of sustained attention seems particularly promising in this respect. However, there is a lack of comprehension on the structural and neural correlates of the behavioral anomalies in these tasks observed in demented and pre-demented subjects. Here we propose a computational whole-brain model of structural (connectivity) and synaptic alterations due to prodromic Alzheimer disease, capable of reproducing the EEG dynamics observed in pre-demented subjects during the execution of a 3CVT task. We recorded EEG activity of 42 Mild Cognitive Impairment (MCI) subjects, 84 Subjective Cognitive Decline (SCD) subjects, and 18 healthy controls (HC) while they performed the 3CVT task. ERP analysis displayed significant differences in both slope and amplitude of the N1 component in the decoding phase of the task, and in the amplitude of the P2 component of the decision phase. Interestingly, all these features were not progressively decreasing with dementia severity, but displayed a larger decrease between the HC and the SCD group than between the HC and MCI group. Of note, this non-linearity was primarily due to a subset of subjects that displayed low levels of functional connectivity (FC) between electrodes. Our computational model reproduced the observed non-linearities and associated them with the degradation of white matter fibers during disease progression. Coherently with experimental results, this degradation led to a second-order phase transition between a low-FC and a high-FC zone: in the low-FC zone, ERP features did not decrease linearly with dementia severity, while they did in the high FC zone. Overall, our model of dementia progression is not only able to reproduce the EEG dynamics observed in SCD and MCI, but also to shed light on the underlying mechanisms.