Resting-state functional and spectral alterations underlying cognitive decline in biomarker-confirmed Alzheimer's disease: A systematic review.

Alzheimer's disease (AD) is characterized by progressive cognitive decline associated with synaptic and large-scale network dysfunction. Resting-state alterations may provide functional correlates of AD pathology, but associations among intrinsic brain activity, biomarkers, and domain-specific cognition have not been systematically integrated across spectral and connectivity frameworks. We systematically reviewed PubMed, Scopus, and Web of Science through May 2025 (PROSPERO CRD42024575913). Thirty studies (N = 5898) examining resting-state EEG, MEG, or fMRI, AD biomarkers, and cognition were synthesized following SWiM guidance. Across studies, resting-state alterations showed converging associations with amyloid and tau measures from CSF, PET, and plasma. Five of six electrophysiological studies reported increased delta/theta activity and three of six reduced alpha/beta activity, supporting neural slowing. Greater low-frequency activity was generally associated with poorer global cognition and episodic memory, whereas reduced higher-frequency activity was linked to poorer attention and processing speed. Among 21 connectivity studies, 10 reported hyperconnectivity and 13 hypoconnectivity, with non-mutually exclusive categories. Hyperconnectivity, particularly within medial and anterior temporal circuits, was more frequent in earlier-stage groups and showed heterogeneous cognitive associations, including poorer and better-preserved performance. Broader hypoconnectivity across posterior default mode, medial temporal-posterior medial, and frontoparietal networks was more prominent in later-stage groups and more consistently associated with poorer global cognition, episodic memory, and executive function. These findings support spectral slowing and stage-related network reorganization as functional correlates of cognitive impairment in biomarker-characterized AD. However, 25 of 30 studies were cross-sectional; therefore, these connectivity patterns should not be interpreted as a demonstrated within-person trajectory.