Nonlinear Modeling of the Dynamic Effects of Arterial Pressure and Blood Gas Variations on Cerebral Blood Flow in Healthy Humans

Adaptive feedback analysis and control of programmable stimuli for assessment of cerebrovascular function

“…The above 3-phase scheme was implemented using an "adaptive inverse approach" [9] with three control states are defined following (1): (3)- (5). The details of the control cases and the parameters used in these equations are provided in Table 1.…”

“…The former maintains cerebral blood flow (CBF) relatively constant despite changes in mean arterial blood pressure (ABP) and the latter reflects the strong effects that changes in PaCO 2 can have on CBF. Many previous studies using system identification techniques to investigate CBF regulatory mechanisms were based on spontaneous fluctuations of ABP (as input) and CBF (as output) to extract information about the dynamic properties of cerebral autoregulation in the frequency or time-domain [3,4,5], and such modeling has also been extended to CO2 reactivity [3,5]. Despite the many advantages of using spontaneous physiological fluctuations in the input and output signals, this approach has several limitations, chiefly the reliability of model based estimates due to poor signal-to-noise ratio and narrow band spectral distributions.…”

Optimising the assessment of cerebral autoregulation from black box models

“…In some of the earlier work [8,13,25,30] the primary concern has been with how well different models fit the data. This In the continuing absence of a gold-standard measure of CA, we take as criteria for assessing performance the ability to distinguish between normal (during normo-capnia) and impaired (during hypercapnia) CA, in healthy adults.…”

“…Even though studies including multivariate models have been published [30,35], so far there has been little analysis as to whether this type of models may offer an improvement in terms of separation between normal and impaired conditions compared to the univariate approaches.…”

“…Most of the studies of CA focus on linear methods [5,[7][8][9][10] with the more recent inclusion of some nonlinear approaches [8,[10][11][12][13]. Although nonlinear techniques can provide improved model fits, their benefit in assessing CA is still unclear with few studies having systematically compared them to linear alternatives.…”