Real-time evaluation of spectral heart rate variability.

The most common spectral indices of heart rate variability are the powers in the high frequency (HF), low frequency (LF) and very low frequency (VLF) bands. These indices have typically been computed using Fourier transforms. In this paper we propose a multirate filtering approach that does not require that the initial R-R intervals be interpolated to a regular sampling grid. The main advantage of this technique is that the HF, LF and VLF indices can now be updated at a high rate, allowing the tracking of short term HRV changes. The key to this real-time evaluation of spectral HRV is the first stage filter that converts the irregularly sampled R-R intervals to a regularly sampled signal.