Normative brain-state trajectories reveal deviation from healthy aging in Alzheimer's disease

Distinguishing healthy brain aging from early neurodegenerative disruption remains a major challenge in Alzheimer's disease. Here, we used resting-state fMRI from the Alzheimer's Disease Neuroimaging Initiative to model large-scale brain-state dynamics relative to a cognitively normal aging reference. A Hidden Markov Model trained exclusively on cognitively normal adults defined latent connectivity states, and a generalized additive model estimated an age-adjusted reference trajectory of transition entropy. Mild cognitive impairment and Alzheimer's disease showed progressively greater deviation from this reference, with the strongest disruption in Alzheimer's disease. A single absolute-deviation score retained much of the predictive information contained in higher-dimensional dynamic features, supporting its use as a compact and interpretable candidate biomarker of altered brain-state organization. These findings suggest that Alzheimer's disease is associated with measurable departure from healthy dynamic brain aging, providing an interpretable framework for future longitudinal and clinically validated studies.