Numerical algorithm for self-consistent stationary level for multidimensional non-stationary time-series

In this paper we consider the self-consistent stationary level of electroencephalogram time series. The practical purpose of this statistics is to construct the disorder indicator. Unlike the classical problem of stationary test of two samples, in our case one should construct an indicator to predict the change in the nonstationary regime. For example, we consider special predictor of an attack of epilepsy.