Optimizing the assessment of autoregulation from black-box models

Autoregulation mechanisms maintain blood flow approximately stable when blood pressure changes. Impairment in autoregulatory function can be identified by black-box modelling with blood pressure as input, and blood flow as output, using only spontaneous variability in both signals. The current paper addresses the issue of how to assess autoregulaton from the model parameters. We propose using a test input in the shape of a band-pass filtered impulse, that reflects the characteristics of blood pressure variations more accurately than the more commonly used step response. Features extracted from the relationship between the new impulse and the best-fit model response are shown to have better dynamic range and stability than a comparable index derived from the step response.