Why am I getting very high values for the very low frequency region of a random signal?

[Joaana]

Hi .. i have attached the front end(prototye) of my project. actually wat i have done is obtained the RR variations (R2-R1) and plotted it. calculated HRV= 60/RR (msec) ....beats per min and plotted it..usd this as input to the PSD.vi (express vi...dB output)... u see the 2 peaks finally (if i use V2/Hz ... i see jus 1 peak)......How do i get result in msec2 and how do i extract the LF and HF power values to calculate LF / HF ratio????

   Plzz help me out.

[griffindore]

Hi Joaana,

From your graph, you see two peaks in the RR variations and the HRV data graph. From your post, it sounds like you are taking the PSD of the HRV data graph and displaying the result in the PSD Auto Regression Model graph. You then see a peak at 0 which signifies the DC offset (or average) beats per minute (roughly 70 from the HRV graph). Then you see another spike at about .37 Hz which signifies the frequency of the two spikes you saw in your HRV graph. You do not see multiple peaks because the two peaks from the HRV data graph were part of the same sinusoidal frequency. Where do you expect to see the LF and HF power values? From first glance you should only have a DC peak and one frequency peak. You could subtract all the values on your HRV data graph by the mean of those values to eliminate the DC offset. This will allow you to see a cleaner PSD Autoregression model in which you may be able to extract a second peak if it exists. With respect to getting a result in msec^2, could you please clarify?

[giora]

Hi Mike,

"Where do you expect to see the LF and HF power values?"

in RR series, hi and low freq can represent different neural processes (hi freq variability, for example, represents parasympathetic activity). usually, the high and low range is predetermined (based on human physiology studies) and not extracted from the data (0.12-0.4 hz, for exmp, is an acceptable band for high freq).

hope that helps

giora

[Amanda_H]

Hi Joaana,

Were you able to use Mike's suggestion for your application? ("You could subtract all the values on your HRV data graph by the mean of those values to eliminate the DC offset. This will allow you to see a cleaner PSD Autoregression model in which you may be able to extract a second peak if it exists.") Let us know; we'd be happy to help!

Have a good day!