Co-activation Patterns Reveal Aberrant Brain Network Dynamics Across Different Stages of Alzheimer's Disease

The mechanisms underlying dynamic changes in brain networks across subjective cognitive decline (SCD), amnestic mild cognitive impairment (aMCI), and Alzheimer's disease dementia (d-AD) remain incompletely understood. This study investigated brain network dynamics across AD stages using co-activation pattern (CAP) analysis of resting-state fMRI data from 62 normal controls (NC), 47 SCD, 60 aMCI, and 55 d-AD participants. We identified seven recurring brain states in the NC group, reflecting activity in high-order cognitive network (HOCN) and primary sensory network (PSN)-HOCN. Analysis of state temporal dynamics, relative to NC, revealed no significant abnormalities in HOCN or HOCN-PSN states in the SCD group. In contrast, persistence disruptions emerged in the aMCI group and progressed to widespread abnormalities across multiple dynamic measures in the d-AD group. As the disease advanced, transition probabilities between states increased while selftransition probabilities decreased, suggesting that the brain gradually lost its ability to maintain stable functional states. In addition, several characteristic state transition pathways consistently emerged across stages, indicating that these temporal dynamics are not random but follow specific functional trajectories. These findings highlight dynamic reorganization in brain networks during AD progression and may inform early diagnosis and interventions targeting functional brain dynamics.