Abstract:

The effect of resampling on spectral analysis of pulse interval series was systematically analyzed. The method of the mean pulse rate as the resampling rate was proposed, which could provide theoretical support for  pulse rate variability. Four methods, the nearest neighbor interpolation, piecewise linear interpolation, piecewise cubic Hermite interpolation and cubic spline interpolation, were applied to resample the simulated pulse interval series. The Welch periodogram and the Autoregressive model were used for spectral analysis with several resampling rates compared. Results showed that the combination of cubic spline interpolation and the Welch periodogram to be an optimal choice for least errors and spectral analysis of pulse interval series. And the mean pulse rate was the optimal resampling rate. Thirty healthy subjects and thirty subjects with coronary heart disease (CHD) were tested to verify the effectiveness of the proposed methods. Results showed that spectral analysis of the subjects with CHD was significantly lower than that of the healthy subjects.

Key words: unevenly sampling, pulse signal, re-sampling, spectral analysis, pulse rate variability