A study of non stationary phenomena of HRV during 24-hour ECG ambulatory monitoring

The frequency components of heart-rate variability (HRV) as evaluated by spectral analysis are now widely used as an index of the autonomic nervous system. Other parameters, such as quantification of asymmetric ascending and descending trends and continuous evaluation of their amplitude, could also be of value. To make these parameters available, the authors describe two new methods: (1) an algorithm to detect the sites of inversion between successive accelerations and decelerations, and to store their wavelengths, numbers and amplitudes; and (2) an adaptive demodulating method based on a sinusoidal function sampled by the RR series. Instantaneous phase and amplitude deviations between the predicted model and the real value give access to the modulation of each frequency studied. For comparison with classical spectral analysis, data were averaged over buffers. Correlations were good for fast and slow variations, but higher for fast ones