ADAPTIVE AR MODELING OF NONSTATIONARY TIME-SERIES BY MEANS OF KALMAN FILTERING

An adaptive on-line procedure is presented for autoregressive (AR) mod eling of nonstationary multivariate time series by means of Kalman fil tering. The parameters of the estimated time-varying model can be used to calculate instantaneous measures of linear dependence. The usefuln ess of the procedures in the analysis of physiological signals is disc ussed in two examples: First, in the analysis of respiratory movement, heart rate fluctuation, and blood pressure, and second, in the analys is of multichannel electroencephalogram (EEG) signals. It was shown fo r the first time that in intact animals the transition from a normoxic to a hypoxic state requires tremendous short-term readjustment of the autonomic cardiac-respiratory control. An application with experiment al EEG data supported observations that the development of coherences among cell assemblies of the brain is a basic element of associative l earning or conditioning.