Dynamic Synergy Network Analysis Reveals Stage-Specific Regional Dysfunction in Alzheimer's Disease

Background: Alzheimer's disease (AD) is a prevalent neurodegenerative disorder characterized by progressive neurodegeneration and connectivity deterioration. While resting-state functional magnetic resonance imaging (fMRI) provides critical insights into brain network abnormalities, traditional mutual information-based methods exhibit inherent limitations in characterizing the dynamic synergistic mechanisms between cerebral regions. Method: This study pioneered the application of an Integrated Information Decomposition (ΦID) framework in AD brain network analysis, constructing single-sample network models based on ΦID-derived synergy metrics to systematically compare their differences with mutual information-based methods in pathological sensitivity, computational robustness, and network representation capability, while detecting brain regions with declining dynamic synergy during AD progression through intergroup t-tests. Result: The key finding are as follows: (1) synergy metrics exhibited lower intra-group coefficient of variation than mutual information metrics, indicating higher computational stability; (2) single-sample reconstruction significantly enhanced the statistical power in intergroup difference detection; (3) synergy metrics captured brain network features that are undetectable by traditional mutual information methods, with more pronounced differences between networks; (4) key node analysis demonstrated spatiotemporal degradation patterns progressing from initial dysfunction in orbitofrontal-striatal-temporoparietal pathways accompanied by multi-regional impairments during prodromal stages, through moderate-phase decline located in the right middle frontal and postcentral gyri, to advanced-stage degeneration of the right supramarginal gyrus and left inferior parietal lobule. ΦID-driven dynamic synergy network analysis provides novel information integration theory-based biomarkers for AD progression diagnosis and potentially lays the foundation for pathological understanding and subsequent targeted therapy development.