SCM

Forum: open-discussion

Monitor Forum | Start New Thread Start New Thread
RE: Short intervals for RMSSD analysis [ Reply ]
By: Aaron Godlaski on 2015-11-20 16:45
[forum:42745]
In general, I would steer clear of a time domain measure like RMSSD given the very short segment length. 30s., even if you have a number of intervals that you intend to average across does not give you much of a distribution to work with. It would be more sensible to use a spectral analysis, and even then extend the length of your segments (if possible). I have not worked with RHRV enough to explore if this is an option though.

RE: Short intervals for RMSSD analysis [ Reply ]
By: Constantino Antonio on 2015-11-20 16:42
[forum:42744]
Dear Dr. Watson,

When analyzing short recordings with CreateTimeAnalysis you should take into account that SDANN and SDNNIDX should not be taken into account (they calculate statistical values using several windows, and in a short recording we only have 1 temporal window). Also, TINN and HRVindex are based in the histogram and thus, they would not yield reliable statistics (in 30 seconds at a normal heart rate we won't have enough points to build a good histogram). The remaining statistics, although computed with just a few points, might be useful for your studies. Thus I would use:

# I assume that data contains 30 seconds of heart rate. I set the size of the window larger than
# the length of the data to avoid unnecessary computations (A warning will be raised, but don't worry)
# Just in case you want to used the histogram-based statistics, I would set numofbins to 10. However, in my opinion, these statistics shouldn't be used.
data = CreateTimeAnalysis(data, size = 60, numofbins = 10)

and then proceed as usual.

I hope this helps.
Kind regards

Short intervals for RMSSD analysis [ Reply ]
By: David Watson on 2015-11-20 16:04
[forum:42743]
We have been using RHRV successfully on a number of recent data sets. It works extremely well. We are now confronted with the need to analyze quite short beat lengths for which we only want a temporal estimate of variability such as RMSSD. I am talking about segments of approximately 30 sec. Is this possible with RHRV? If so what adjustments need to be made to the parameters to optimize success of such an attempt.
Best regards
David

Thanks to:
Vienna University of Economics and Business Powered By FusionForge