Data-driven and data-assimilation approaches to individual human brain dynamics
Recently, we proposed a data-driven method for analyzing the rhythms in the metastable human brain. The method can label the metastable rhythmic dynamics as multiple torus attractors. Then, we showed experimental evidence of the metastable two-dimensional torus attractors associated with the autistic-like traits for the resting human brain. Further, evidence that each attractor can generate two kinds of the coupling dynamics, i.e., the phase-phase coupling (PPC) dynamics for delta-band oscillations and the phase-amplitude coupling (PAC) dynamics for delta- and alpha-band oscillations, has been obtained where different PPC states can characterize different PAC states. Motivated by this experimentally found PPC-PAC relationship, in this study, we focused on modeling the individual delta-band phase dynamics in the resting human brain by using the Kuramoto model with the uncertainty by a data assimilation (DA) approach based on our previous data-driven approach. We estimated the parameters that can characterize the PPC connectivity from the observed delta-band phase dynamics by using the four-dimensional variational method (4D-Var), and then the Kuramoto model was individualized. In this presentation, we show the possible effectiveness of the DA study inspired by the previous data-driven study towards further elucidation of the individual human brain dynamics.
Dr. Takumi Sase RIKEN Center for Brain Science