How to check steady-state condition from cardiovascular time series

In this study we propose a procedure to automatically check the hypothesis of steady state from the dynamics of the cardiovascular signals, based on the 'run test'. The total number of runs r(tot) is computed from the series of systolic and diastolic blood pressures and R-R intervals. If r(tot) is lower than a given threshold, the steady-state hypothesis is rejected. Because of long-term correlations affecting the cardiovascular series, the thresholds at the 5% significance level were estimated calculating the rtot distributions from a public database of cardiovascular signals collected in steady-state conditions. The procedure was applied to quantify the steadiness of baseline recordings, and to identify sub-periods of steady state during a sequence of physical activities. Results showed the capability of the procedure to automatically detect steady-state conditions, and to identify when the steadiness is lost because of disturbing factors or transients occurring during the recording.