Fig. 2: Neural fragility and neural excitability in patients with AD and age-matched controls.

Neural excitability (aperiodic spectral slope) in healthy older adults was highest in the medial and dorsal cortices of frontal and temporal lobes (A), while neural fragility (linear dynamic instability) was greater in the dorsolateral than medial cortices (B). An LMM analysis showed no associations between neural excitability and neural fragility in the healthy brain (C). AD patients retained the same spatial gradient distributions for neural excitability and neural fragility as in controls but showed increased values in both metrics (D, E). In AD patients, higher neural excitability correlated with higher neural fragility (F). The regional patterns of increased neural excitability and neural fragility in AD patients vs. controls involved distinct anatomical regions (G, H). In a linear mixed model (LMM) with repeated measures, patients with AD showed higher neural excitability (I; t = 3.35, P = 0.0011) and higher neural fragility (I; t = 4.08, P < 0.0001), compared to controls. Global cognitive decline in patients with AD was associated with both increased neural excitability (J) and neural fragility (K) The X axes of J and K show the four quartiles of MMSE. The markers depict the least square means corrected for age derived from LMM analyses. In C and F, the scatter plots indicate each subject’s data from 210 cortical regions; the black lines denote the model predictions; the shaded areas indicate the 95% confidence intervals of the model prediction. Brain renderings in G and H show statistical significance from group comparison between AD and controls after covarying age and thresholded at FDR 5%. The subplots I, J and K denote least square means with the error bars indicating standard error. AD Alzheimer’s disease, FDR false discovery rate, LMM linear mixed model, MMSE Mini Mental State Exam (scores range from 0 to 30, with higher scores denoting better cognitive function).