Determining aging trajectories though hyperbolic embeddings of MEG brain networks

Background: Investigating age-related changes in MEG brain networks offerssignificant potential for comprehending aging trajectories and unveiling anomalouspatterns associated with neurodegenerative disorders, such as Alzheimer’s disease. Inthis study, we extended a deep learning model called Fully Hyperbolic Neural Network(FHNN) to embed MEG brain connectivity graphs into a Lorentz Hyperboloid model forhyperbolic space. Through these embeddings, we then explored the impact of aging onbrain functional connectivity across multiple decades.Method: We analyzed data from 587 participants enrolled in the Cambridge Centrefor Ageing and Neuroscience (Cam-CAN) longitudinal study. Notably, we introduceda unique metric—the radius of the node embeddings—which effectively proxies thehierarchical organization of the brain. We leveraged this metric to (i) assess whetherwe can decode age-related information, and (ii) characterize subtle hierarchicalorganization changes of various brain subnetworks attributed to the aging process.Result: Our decoding results revealed that the hyperbolic radius carries substantiallymore age-related information compared to all other conventional graph-theoreticmeasures examined, underscoring the effectiveness of employing hyperbolicembeddings to characterize the aging process. An examination of hyperbolic radiusalteration patterns across decades exposed numerous subnetworks showcasing adecline in hierarchy during aging, with some displaying gradual changes and othersundergoing rapid transformations in the aging brain (illustrated in Figure 1cde).Conclusion: Overall, our study presented the first evaluation of hyperbolic embeddingsin MEG brain networks, introduced a novel measure of brain hierarchy, and used thismeasure to highlight aging trajectories in the large cohort of the Cam-CAN dataset.A prominent finding was the reduction of hierarchy across a substantial number ofsubnetworks throughout the aging brain. This hierarchy reduction could imply a shiftin the brain network configuration impairing cognitive processes.This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, providedthe original work is properly cited.© 2024 The Alzheimer’s Association. Alzheimer’s & Dementia published by Wiley Periodicals LLC on behalf of Alzheimer’s Association.Alzheimer’s Dement. 2024;20(Suppl. 2):e086649. wileyonlinelibrary.com/journal/alz 1 of 2https://doi.org/10.1002/alz.086649