Optimal Window for the Estimation of Very Low Frequency Content in Heart Rate Variability Analysis

摘要

The spectral analysis of heart rate variability (HRV) is an accepted method for assessing the autonomic control of the cardiovascular system. The electrocardiographic recordings used to extract the time interval between consecutive R waves (RR intervals or tachogram) should last 5 min or 24 h, according to the guidelines published in 1996 by the European Heart Journal (TASK FORCE). The three frequency bands recognized are the VLF (0.003-0.04 Hz), the LF (0.04-0.15 Hz) and the HF (0.15-0.4 Hz), which are associated with cardiovascular modulatory mechanisms. Given that the estimation of VLF in 5 min recordings is unreliable and that in some circumstances (orthostatism, controlled breathing, etc.) it is not possible to obtain 24 h recordings, it becomes necessary to consider other window sizes in order to estimate it with greater certainty. To show how the size of the window affects the estimation of the power spectral density, synthetic signals were evaluated using the Welch periodogram, comparing the power and spectral resolution obtained with conventional 5 min windows versus results obtained with windows that ranged from 300 to 3000 s in steps of 300 s, plus an additional one of 4000 s. Noise-free signals were generated, contaminated with White Gaussian Noise and mounted on a linear trend, to approximate the conditions of a real tachogram. The results suggest that the optimal size of the analysis window is 50 min, decreasing the power estimation error of the VLF band from 24.91 to 8.06%, increasing the spectral resolution and increasing the confidence in the evaluation of the three frequency bands defined for the HRV, especially for the VLF. These results suggest an alternative analysis for recordings with duration less than 24 h but that require evaluating the VLF, as in the HRV recordings of patients during their haemodialysis session.