Nonlinear multivariate analysis of dynamic cerebral blood flow regulation in humans

The dynamic relationship between cerebral blood flow, arterial blood pressure and arterial CO₂ is studied using the Laguerre-Volterra network methodology for modeling multiple-input nonlinear systems. Spontaneous beat-to-beat cerebral blood flow velocity and mean arterial blood pressure fluctuations, as well as breath-to-breath end-tidal CO₂ fluctuations are analyzed and the Volterra kernels of the system are obtained. It is found that, while pressure changes explain most of the blood flow velocity variations, the inclusion of end-tidal CO₂ fluctuations as an additional input variable can improve the prediction accuracy of the model output considerably. The model includes also nonlinear interactions between pressure and end-tidal CO₂ and their impact on cerebral blood flow.