Cameron Craddock1, Stephen M. LaConte1
We propose a method for deriving functional connectivity maps using multivariate prediction analysis regression. This method provides accurate estimation of the time course of activity for a resting state network (RSN) of interest from a never-before-seen dataset. This approach is evaluated for 10 RSNs on a resting state test-retest dataset acquired from 26 subjects. The proposed method is able to accurately estimate RSN activity when at least 5 minutes of data are available for training. This method provides a framework for tracking RSN activity in real-time as well as comparing methodological tradeoffs inherent in resting state functional connectivity analyses.