1999 Volume 11 Issue 5 Pages 848-854
The stationarity of time series is assumed in many methods of time series analysis such as the chaotic time series analysis. Thus, it is important to analyze the stationarity before applying such methods to observed time series. We analyze the stationarity of respiration time series of newborn infants with KM_2O-Langevin equations. We apply the analysis to long time series of preterm and term infants. At first the local stationarity within short time windows is investigated, then the global stationarity of the entire time series is estimated on the basis of the result of the local stationarity analysis. Our numerical results show that the local stationarity decreases in parts where the amplitude suddenly varies as typically observed in prolonged apnea, and that the global stationarity of the time series of preterm infants tends to be lower than that of term infants.
