Evaluation of pulse rate variability obtained by the pulse onsets of the photoplethysmographic signal

This work presents the evaluation of pulse rate variability (PRV) obtained from pulse onsets of photoplethysmographic (PPG) signals. Three published algorithms were used to determine the pulse onsets: diastolic point, maximum second derivative and tangent intersection. Temporal series of pulse onsets were obtained for each method, and several variability indices were derived from these series. Simultaneous ECG and PPG records were acquired from 37 healthy volunteers to evaluate the interchangeability between PRV indices and heart rate variability (HRV) indices by the Bland-Altman method. Furthermore, the concordance correlation coefficient was used to correlate the indices. In all the cases, PRV indices obtained through the tangent intersection method showed better accuracy and precision (Bland-Altman analysis, bias ± 1.96 standard deviation: low frequency, LF(ms)(2) = -28.06 ± 72.68; high frequency, HF(ms)(2) = -68.23 ± 192.85; high frequency in normalized units, HF(nu) =-2.02 ± 7.08; LF/HF = 0.17 ± 0.71) and higher correlation (concordance correlation coefficients: low frequency, LF(ms)(2) = 0.99; high frequency, HF(ms)(2) = 0.98; high frequency in normalized units, HF(nu) = 0.97; LF/HF = 0.90) with HRV indices than other methods, and could be used as a good surrogate of HRV.