Application of the Lomb-Scargle Periodogram to InvestigateHeart Rate Variability during Haemodialysis

Figure 5

PSD estimation via the LS periodogram using the same original data series and symmetric RR interval selection (±10%), but with five different methods of editing dubious RR intervals. Methods 1 (“baseline”), 2 (“rules”), 4 (OUGP), and 5 (EMD) are considered successful as they were able to identify the LF and HF peaks corresponding to the sine waves in (3). Methods where dubious RR intervals were replaced by cubic spline interpolation, linear interpolation, or by the mean average of one minute of data are considered unsuccessful as background peaks are emphasised and HF peaks are lost in the noise. FAP levels are indicated by the three horizontal lines on each graph, with the top (purple) line showing a FAP of 1%, the mid (yellow) line showing a FAP of 10%, and the lower (orange) line showing a FAP of 50%. (a) Lomb–Scargle baseline. (b) Rules edited RR intervals. (c) Cubic spine interpolation. (d) Linear interpolation. (e) Mean average replacement. (f) OUGP. (g) EMD.

(a)

(b)

(c)

(d)

(e)

(f)

(g)