[The spectral analysis of heart rate variability. A comparative study between nonparametric and parametric spectral analysis in short series].

OBJECTIVE

to compare parametric (AR) and non parametric (FFT) spectral analysis results obtained in 512 beats series.

INTERVENTIONS

104 healthy subjects with normal physical examination and electrocardiogram were studied. The Ecg was recorded at rest, with controlled breathing at 15 cycles/min., and sampled at 300 Hz. The spectral VLF, LF and HF were calculated with FFT algorithm. For the same series, an auto-regressive analysis (AR) with optimized choice of the order of the model (AIC criterion) have been computed, VLF, LF and HF components were identified by AR spectral decomposition.

RESULTS

In both groups, athletes and sedentary, there were no statistically differences between VLF, LF, HF and LF/HF spectral indices computed by the two methods.

CONCLUSION

the results suggest that with controlled breathing it does not seems to exist any advantage in the use of AR spectral analysis to compute spectral components of heart rate variability, which is much more laborious that fixed bands non parametric FFT analysis.