Abstract
According to the AT(N) framework, Alzheimers disease (AD) can be biologically characterised by amyloid
deposition, tau pathology, and neurodegeneration, yet how these processes relate to whole-brain
dynamics remains unclear. Here, we investigated these relationships and the link between whole-brain
dynamics and cognitive dysfunction using multimodal neuroimaging from 145 ADNI3 participants. We
combined multiscale measures of turbulent dynamics with responses to in silico perturbations obtained
from whole-brain models. AD progression was characterised by a scale-dependent reorganisation of
turbulent dynamics where information flow showed the strongest contribution to the dynamics-related
variance in cognitive performance. Large-scale turbulence was associated with all three AT(N) components.
Model-based information capability and susceptibility measures progressively decreased across
AD stages and were associated with amyloid burden and regional tau deposition. Model-free features
discriminated comparisons involving AD dementia, whereas adding model-based measures produced
modest, metric-dependent changes in classification performance. Exploratory analyses showed spatial
alignment between large-scale turbulence alterations and AD-risk gene-expression gradients, with
APOE among the strongest associations in both disease-stage contrasts examined. Lastly, we found a
progressive decoupling between turbulent dynamics and the E:I balance across AD stages. Together,
these findings link model-based and model-free measures of whole-brain dynamics to proteinopathy,
neurodegeneration, and cognitive impairment, supporting their complementary value for characterising
AD progression.