CMStatistics 2018: Start Registration
View Submission - CMStatistics
B1251
Title: Detecting regime changes in community structure of brain networks using dynamic stochastic block models Authors:  Chee Ming Ting - King Abdullah University of Science and Technology (Saudi Arabia) [presenting]
Siti Balqis Samdin - King Abdullah University of Science and Technology (Saudi Arabia)
Hernando Ombao - King Abdullah University of Science and Technology (KAUST) (Saudi Arabia)
Abstract: Brain networks exhibit the property of modular community structure with highly inter-connected nodes within a same module, but sparsely connected between different modules. Recent neuroimaging studies also suggest dynamic changes in brain connectivity over time. We present a statistical approach based on dynamic stochastic block models (SBM) to characterize changes in community structure of the brain functional networks inferred from neuroimaging time series data. The dynamic SBM is a non-stationary extension combining a static SBM with a Markov process to allow for temporal evolution of the community membership of nodes and the network connectivity. The model is formulated into a state-space form with sequential estimation of the time-varying parameters by Kalman filtering. We further partition the time-evolving community structure into recurring, piece-wise constant regimes or states using an infinite hidden Markov model that can learn an unknown number of states from the data. The method is applied to resting-state and task-based functional magnetic resonance imaging (fMRI) data to detect dynamic reconfiguration of network structure of the brain.