Non-linear modeling of cerebral autoregulation using cascade models

Abstract

Autoregulation mechanisms maintain blood flow approximately stable despite changes in arterial blood pressure. A model that characterizes this system is of great use not only in understanding cerebral hemodynamics but also for the quantitative assessment of function/impairment of autoregulation. Using arterial blood pressure (ABP) as input and cerebral blood flow velocity (CBFV) as output, the autoregulatory mechanism was modeled using only spontaneous variability in both signals, in accordance with previous work. In this study a non-linear approach, based on a cascade, also known as block structure models, is presented, whose parameters are estimated by Differential Evolution. The results were compared with other linear and non-linear approaches previously used to model cerebral autoregulation. The performance of each model was assessed by the model’s predicted CBFV in terms of the normalized mean square error (NMSE) and the correlation coefficient. The results show that for relatively short signals (150 sec) containing only spontaneous fluctuations, cascade models performed better than a frequency domain method but are not significantly different from linear time-domain techniques tested. These results also show that slightly better performance can be obtained with the cascade models compared with more complicated non-linear models with the advantage of having more easily interpretable parameters and a simpler structure that facilitates their use in diagnostic methods.