Controllability of the brain system has been studied in the domain of network science. However, most of studies on the controllability of the brain have not considered the nonlinear nature of the brain. In the present study, we suggest a computational framework to control the brain system with a consideration of nonlinear brain dynamics. Our framework is based on a hypothesis that a brain with a disease has specific brain dynamics different from that of a normal brain and can be analyzed by using an energy landscape analysis. For both of normal and abnormal brain systems, multistable activation states (attractors) and transition rates were investigated by performing an energy landscape analysis based on a pairwise maximum entropy model. In the current virtual framework, we simulated how dynamics of a disease brain can be changed to that of the normal brain by external treatments under biological constrains. By doing this, we tried to find a strategy for optimal treatments that control the target brain to generate brain state dynamics similar to that of the healthy brain. We assumed that the target brain changes not only at a treated region or treated connectivity, but also it induces changes in the neighbors that the treated region interacts. By allowing changes in the neighborhood in response to the treatment to a target region, we showed an optimal controllability that takes into account of the nonlinear responses of the brain after treatment. We expect that this computational framework for controllability would help treatment planning for the nonlinear brain system, after empirical evaluation and validation.
Acknowledgments
This research was supported by Brain Research Program and the Korea Research Fellowship Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science and ICT (NRF-2017M3C7A1049051 and NRF-2017H1D3A1A01053094).