Objective: Cerebral autoregulation is commonly assessed through mathematical models that use non-invasive measurements of arterial blood pressure and cerebral blood flow velocity. There is no agreement in the literature as to what is the minimum length of data needed for the cerebral autoregulation coefficients to stabilise.
Approach: We introduce a simple empirical tool for studying the minimum length of time series needed to parameterise three popular cerebral autoregulation coefficients ARI, Mx and Phase (in the low frequency range [0.07-0.2] Hz), which can be easily applied in a more general context. We use our recently collected data, from which we select high quality (absence of non-physiological artefacts), baseline ABP-CBFV time series (16 min each). The data were beat-to-beat averaged and downsampled at 10 Hz.
Main result: On average, ARI exhibits greater variability than Mx and Phase, when calculated for short intervals; however, it stabilises fastest.
Significance: Our results show that values of ARI, Mx and Phase calculated on intervals shorter than 3 min (1800 samples), 6 min (3600 samples) and 5 min (3000 samples), respectively, may be very sensitive to changes in the length of data interval.