Deeper Dive Into HRV

After I did my Heart Rate Variability measurements (check out my earlier post, if you haven’t), I suspected that RMSSDs had wide range. For example, when I compared my RMSSD just before the long run and 45-hours after the long run, I ended up getting 64% higher number. This led me to suspect that RMSSD varies quite a bit depending on multiple factors.

So I calculated RMSSDs using different number of heart beats. Earlier I used 336 heart beats (about 5 minutes worth of heart beats) to measure my HRV. (Note that these are the data that I collected between December 28th and 30th, 2015; the same set of data that I discussed on my earlier post.)

  • Measure 1: 25 minutes before the run
    • 60 Heart Beats: 50.64 msec
    • 100 HBs: 40.962 msec
    • 336 HBs: 25.551 msec
  • Measure 2: 1 hour 45 minutes after the run
    • 60 HBs  : 5.534 msec
    • 100 HBs: 5.196 msec
    • 336 HBs: 4.505 msec
  • Measure 3: 23 hours after the run
    • 60 HBs  : 17.606 msec
    • 100 HBs: 20.348 msec
    • 336 HBs: 14.396 msec
  • Measure 4: 45 hours after the run
    • 60 HBs  : 93.512 msec
    • 100 HBs: 73.093 msec
    • 336 HBs: 42.032 msec

Looking at these numbers, it is easy to see that HRVs are not about exact measurement, but instead it’s about relative scale. For example, Measure 4 could have more than twice as large HRV if I took the first 60 HBs only as opposed to looking at the 336 HBs.

Why? Because when the heart rate had other reasons to vary, such as external stress or change in posture, HRV would be greater to include the increase in heart rate.

In fact I think what might have happened on Measure 4 was exactly that. While hurrying to take the RR recordings, I got up, got my chest monitor on, and lied back down on the bed quickly to start the measurement. That explains why the heart rate was dropping from 80.

The below graph shows my heart rate changes on Measure 4.


So the first point to note is that it’s important to minimize the external factors that may change the heart rate. I think the best way is to lie down for a few minutes before the measurement, and stay lying down during the measurement.

The other point is that HRV should be thought of as a range. From multiple measurements I took, it looks like there are three ranges that I could identify for myself:

  1. Relaxed: 25 msec or greater
  2. Recovering: 10 – 24 msec
  3. Stressed: 9 msec or less

It would make sense to compare 50 msec (60 HBs) and 25 msec (336 HBs) in Measure 1 to conclude that I was under physical stress (because I was not).

What I do not know: How much psychological stress would affect my HRV, and how long the effect would be sustained in my HRV. I’m sure experimental psychologist would have many sample points in this area.

So what’s the bottom line?

HRV definitely works in measuring the physical stress, but to properly measure I’ll need to lie down and spend at least 5 minutes of measurements to get a relative idea of how stressed I am.

I wonder if there are any experts who can share their findings. Please feel free to chime in.


Heart Rate Variability

[UPDATE 12/30/2015] While going over the additional data, I realized that I had a typo in Excel formula while computing RMSSD (Root Mean Square of Successive Differences). As a result, instead of average of all RR intervals, I was using average of only the first and last RR intervals. Oops. So I updated the RMSSD numbers and my concluding comment.


One of the features of Polar V800 is RR Recording. It measures heart rate variability (HRV). Heart rate variability is the differences in consecutive heart rates. For example, if heart beats at t=0, 1, 1.9, and 2.85 second, the heart rate would be 60, 66.7, and 63.2 beats per minute, and the HRV would be 6.7 and -3.5 bpm.

It’s also often expressed as RR-intervals, which is short for R-wave-to-R-wave interval. R-wave is the wave created by heart beat where the peak is found. RR-interval is also known as Inter-beat Interval (IBI). In the above example, RR-intervals (or IBIs) would be -0.1 and 0.05 second. (For more information on R-wave, visit here.)

It is nothing but a calculated value from straight off heart beat measurement. So what’s the deal? Why would anyone bother measuring their HRV?

It turns out that the HRV has been used in psychology experiments to represent how relaxed the subjects are, and how ready they are to deal with external stresses. The greater the HRV is (meaning the greater the heart rate varies from second to second), the more relaxed the subject is. On the other hand, the smaller the HRV is (i.e. the heart rate is more or less constant), the more stressed she is.

If anyone cares to recall their human anatomy class, HRV measures whether sympathetic or parasympathetic nervous system is kicking in. Sympathetic nervous system has to do with fight-or-flight response. It increases heart rate, and dilates pupil among other stress-related physical responses. Parasympathetic nervous system has to do with slowing down physical responses, such as decreased heart rate, stimulating digestive system, etc.

Because the HRV provides a simple measurement of how relaxed and ready we are, some athletes are using it to see whether they are over training. Thinking is that if the HRV shows little variation, we are under stress, and may not be the right time to start the training.

When I found out about HRV, it seemed counter intuitive. I would have thought that if I’m relaxed, I would see little variation on my heart rate.

Another thing that seemed odd to me was that I couldn’t find a scale (even relative scale) of what typical HRV range would be.

So I decided to experiment.

Minutes before my 15-mile run yesterday, I took my HRV measurement to establish my baseline. Not knowing what to expect, I wanted to get a good sample size, and I figured 5+ minutes would have enough sample points (about 300+ heart beats). Also to minimize the effect of other stresses, I took the measurements while lying down.

Here’s what my RR intervals looked like before my run.


And here’s what they looked like an hour and 45 minutes after my 15-mile run.



As predicted by literature, my heart rate varied noticeably more before my long run, compared to the one after my long run.

To quantify the variability, I ran the Root Mean Square of Successive Differences (RMSSD) of the 337 heart beats. (When I was rested, I had about 337 heart beats in 5 minutes.) BTW, if you want to find out about how to compute RMSSD, you can reference this article that I found.

And to measure how fast I was recovering, I took two additional 5-minute measurements, 23 hours and 45 hours after the run, and computed the RMSSD.

  • RMSSD 00:25 before 15-mile run: 25.551 msec
  • RMSSD 01:45 after 15-mile run   :    4.505 msec
  • RMSSD 23:00 after 15-mile run   : 14.396 msec
  • RMSSD 45:00 after 15-mile run   : 42.032 msec

It looks like my HRV drops to about 18% after my long run. Surprisingly I seem to get full recovery of HRV (plus some more!) within 45 hours of the long run!

That’s way faster than what my V800 estimates as my long run recovery time.

Oh well, even if my heart may be willing, my sore feet say I still needed a rest today. 🙂