Time series analysis in endocrinology.
Matthews DR.
Many hormones are secreted in pulses or oscillations. If deductions about amplitude or frequency of these oscillations are to be made, then the oscillatory attributes must be analysed. The method of doing this is known as time series analysis. Samples for such analysis need to be properly spaced, taken for a sufficient period of time and de-trended. The techniques of pulse counting, autocorrelation or Fourier transformation may then be applied to demonstrate dominant features in groups of subjects. The advantages and disadvantages of these methods are discussed.