A new paper by Dominik Krezeminski is published on Network Neuroscience. We extended an energy landscape method to quantify the occurrence probability of network states in resting-state MEG oscillatory, which was derived from a pairwise maximum entropy model (pMEM). The pMEM provided a good fit to the binarized MEG oscillatory power in both patients with juvenile myoclonic epilepsy (JME) and controls. Patients with JME exhibited fewer local minima of the energy and elevated energy values than controls, predominately in the fronto-parietal network across multiple frequency bands. Furthermore, multivariate features constructed from energy landscapes allowed significant single-patient classification. Our results highlighted the pMEM as a descriptive, generative, and predictive model for characterizing atypical functional network properties in brain disorders.
The paper is now available online.