Kalman filter modeling of cerebral blood flow autoregulation

A parameter estimation scheme for dynamic systems is employed to simultaneously estimate the states and parameters of the model of human cerebral blood flow velocity as a function of mean arterial blood pressure. The estimation results show 20-40% reduction in the output mean square error compared to that of the one obtained from the computer model addressed in a paper by Ticcks, et al. (1995). The estimation scheme estimates the parameters and states of the system, as well as the level of the observed and process noise variances. This approach is more extensive than the one that was applied to the same system in the previous work by Kamangar, et al. (2002), in which only the Kalman filter was applied and the system was restricted to some specific constraints.