A nonstationarity test for the spectral analysis of physiological time series with an application to respiratory sinus arrhythmia.

Applied a nonstationarity test for use in the spectral analysis of physiological time series to data obtained in a typical psychophysiological setting from 41 5-yr-old, 46 7-yr-old, 36 9-yr-old, and 11 adult males under 4 conditions: rest, paced breathing, vigilance, and reaction time (RT). Most cardiac sequences were nonstationary. However, for every S stationary stretches could be found that were long enough to apply spectral analysis. The comparison between stationary data and data not tested for stationarity indicated that nonstationarity affects the spectral statistics that are computed in spectral analysis of cardiac time series. A finding of significant interactions of stationarity, experimental condition, and frequency band for coherence and phase spectra suggests that nonstationarity may constitute a confounding factor in any significance test of the effect of task manipulations on these spectral indices.