Biograph: Wenbin Shi, PhD, is an assistant professor at Beijing Institute of Technology. After graduating from Beijing Jiaotong University, she did her postdoctoral research at Tsinghua University from July 2016 to July 2018. She was a visiting scholar at Harvard university during 2014-2015. Her expertise is electrophysiological signal analysis and processing, complex system theory and its application in sleep medicine, interested in techniques to process electro-biological signals and to promote clinical applications of complex physiological dynamics. At present, she has published more than 20 international peer reviewed papers in IEEE Transactions on Biomedical Engineering, Communications in Nonlinear Science and Numerical Simulation, Scientific Report, Physica A, Nonlinear Dynamics etc., with one authorized patent. She is the reviewer of IEEE Journal of Biomedical and Health Informatics, IEEE Transactions on Neural Systems and Rehabilitation Engineering, Physica A, etc.

Title: measuring multiscale complexity in human sleep electroencephalography

Abstract: Increasing reports indicated that multiscale fashion is a general and efficient approach to investigate various natural and physiological states. To this end, a generalization of the q-complexity-entropy curve in the multiscale fashion (GMCE curve) is introduced. In addition, the concept of the minimum permutation entropy and the maximum statistical complexity are proposed to simplify the visualization of entropy/complexity in the multiscale fashion, so-called generalized multiscale permutation entropy (GMPE), which is proposed to yield a spectrum of entropy and complexity. We demonstrate that the proposed GMCE curve or the GMPE method is capable of identifying an irregular oscillation is either stochastic or chaotic, highly predictable or long-term correlated. Simulations include Gaussian white noise and 1/f noise, stationary and fractionally integrated autoregressive processes, logistic map and Hénon map. In the application of the GMCE curve and the GMPE method in real datasets, especially in the sleep analyses, both the minimum permutation entropy and the maximum statistical complexity show significant differences across sleep stages (p < 0.0001*), whereas all five sleep stages are differentiable in the multiple comparisons. Both the real datasets present clearer visualization and discrimination among basin imperviousness or sleep stages than the standard multiscale entropy. Our approach enables us to investigate irregular oscillations with generalization in timescales and Tsallis q-entropy.