@Tlinlo7don it's not only the sample rate but what the device is looking for which is R-R intervals. For the optical sensor, it's a lot more complicated to find the R-R interval than for the ECG device. The R wave is the biggest wave generated during normal conduction in your heart and simply speaking, detecting the R wave indicates a beat. An optical sensor looks at the blood flow, not the heart. If you look at the data obtained by the PolarH10 strap it detects up to 5 R-R intervals per second. This is enough to find N-N beats (for computing SDNN) and provide accuracy for R-R intervals (even during exercise). The set of R-R intervals may look like this:
time: (1.102, None, None, None, None) [s]
time: (1.006, None, None, None, None) [s]
time: (1.022, None, None, None, None) [s]
time: (1.05, None, None, None, None) [s]
time: (1.123, None, None, None, None) [s]
time: (1.12, None, None, None, None) [s]
time: (1.234, None, None, None, None) [s]
Usually, in rest lower sampling rate does the job but another example (from my 5k race) could be this:
time: (0.353, 0.354, 0.353, None, None) [s]
time: (0.357, 0.354, 0.356, None, None) [s]
time: (0.355, 0.357, 0.357, None, None) [s]
time: (0.356, 0.358, None, None, None) [s]
time: (0.36, 0.358, 0.356, None, None) [s]
time: (0.363, 0.361, 0.36, None, None) [s]
This shows that sampling 5 times per second does the job (I have never seen all the samples filled). This is enough to calculate RMSSD but here's the catch. RMSSD and SDNN are very sensitive to data artifacts and this happens to ECG devices, too. In general, it isn't simple to filter out artifacts, and sometimes despite suspecting a signal to be an artifact it's better to leave it because in the short term it won't have much of an effect on overall HRV. I usually filter it by looking at any unusual spike caused by a single beat (anything that is not falling into the N-N beat category). Simplifying, the filter checks the difference between the last and current interval looking for such sudden spikes, and applies a threshold. The point is, sometimes there is no way to know for sure that the beat is a glitch.
With R-R intervals I can build a beats timeline and put the HR on it so can see how it changes in time. It is true, that the higher HR than the lower intervals are but RMSSD doesn't follow that rule. Just by shifting a single beat in 1min sample, the RMSSD may jump from let's say 38ms to 120ms and a different change will change it to 8ms. This metric is very data sensitive and I usually try to run rolling RMSSD on my data using the 5-10sec domain. Each domain that is suspected to contain an artifact is verified and either filtered out or included. It's not rocket science and my results are aligned with EliteHRV, and easy to verify. I've been researching long-term HRV and for that SDNN seems to be a far better metric than RMSSD. The SDNN is the one for medical evaluation of cardiac risk and it predicts both morbidity and mortality ( so, @IronChuck, it's not just about stress levels 🙂 but in fact actual cardiovascular health including the prediction of SCD in athletes and risk of death from epilepsy - there is a lot more research going on).
The problem with HRV is that this metric isn't made for regular people. Yes, it is the nice selling point of the device but there are very big brains trying to understand what's the use of it and despite the general rule that high HRV = good, and HRV = bad it isn't that simple. The first thing about HRV should be making sure that the data is accurate, then verifying the methodology. We don't know the time domain for the data, we don't know how the artifacts are treated, and we don't know how RMSSD is applied through such a long set of data. Maybe what we get is a snapshot of the last 5 minutes before detecting waking up (which would make sense for morning readiness!)? Maybe not? Nobody knows so from my point of view, the data isn't relevant. I can't tell how it's obtained, how it's processed, and how it's interpreted. As well I could get a number drawn in a lottery and it would have the same meaning to me.
Best Answer
