UPenn's brain 'epidemic' model predicts Alzheimer's tau spread from neuronal activity

Neurodegenerative diseases like Alzheimer's are increasingly understood as spreading processes, where misfolded proteins propagate across anatomically connected brain regions. Prior network models treated this as passive diffusion, ignoring the experimental evidence that neuronal firing actively promotes protein transmission. To close this gap, Alexandersen, Kulkarni, Davis, Roemer-Cassiano, Franzmeier, and Bassett coupled a general node-activity process to susceptible-infected-susceptible (SIS) dynamics on multiscale brain networks. Their theoretical framework derives an epidemic threshold that separates benign seeds from pathological growth, and shows how neuronal activity shifts this threshold and redirects spreading by mixing structural network modes. They further decompose the effect into regional mean activity and within-region variation, capturing heterogeneity that brain imaging cannot directly resolve. Stochastic simulations confirm the predictions across synthetic networks.

The team then validated the model against longitudinal human positron emission tomography (PET) data, using regional glucose metabolism as a proxy for neuronal activity and tau accumulation as the disease marker. Adding activity to the network model reproduced spatial patterns of tau progression that standard structural connectivity or established disease markers could not explain. Across individuals, the model's predicted epidemic threshold also correlated with how broadly pathology spread through the brain. These findings connect epidemic theory to neurodegeneration, position neuronal activity as a driver of Alzheimer's progression, and suggest that activity-modulating therapies could slow or prevent pathological spread. The work opens a new computational lens on prion-like protein propagation, with potential implications for personalized prognosis and treatment timing.