Are accuracy studies of consumer grade heart rate monitors reproducible?

Last Updated on August 17, 2023 by pg@petergamma.org

We have published here a table of accuracy values for different heart rate monitors:

These are all values from different papers which use different protocols and methods. The data in this table cannot be compared to each other. And what stroke us: We have heart rate accuracy testers like DC Rainmaker and Rob ter Horst which test heart rate accuracy on a regular basis with several 10 000 of views on YouTube, which make a big effort for accuracy tests. But for whom are these accuracy tests helpful? Do they have any consequences for someone?

If we look at the papers we listed, is there a single paper among these papers which uses a standartized protocol which is also used in a second paper? A protocol which is used on a regular basis to test new heart rate monitors which come to the market? We do not know of such a paper which offers that.

What is striking is the big diversity of testing protocols in different papers:

Does this mean that all the authors of these papers gave up the topic accuracy studies of these devices? Is it not possible to reproduce their papers? To write a second paper with the same protocol, but a new watch?

We do not know how it is with the signal quality studies as those which have been done for the Polar H10:

https://pubmed.ncbi.nlm.nih.gov/31004219/

We did not study in detail this papers and papers about the Polar chest straps from the Polar H 1 chest strap to the H10 chest strap, since chest straps are not suitable for our application.

Also if we compare the methods which are used in the above papers we mentioned another thing stroke us. Sports scientists looked at individual peaks after the recordings to get data from those to calculate the signal quality. But cardiologists from the Cleveland Clinic in Ohio used in a similar paper where they studied the accuracy of the Polar H7 and the Apple watch III very expensive 3 lead ECG reference devices, stood next to the treadmill and looked at every single ECG peak by by eye during the activity:

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6732081/

Is this not the highest accuracy level which is possible? Are there not differences in the quality of methods cardiologists use and sports scienitsts use? Cardiolgists do clinical studies where every single ECG peak matters, and sometimes it is about live or dead. Do sports scientists make the same big effort as cardiologists in their accuracy studies? For instance in the study with the Polar H10 chest strap we mentioned above? Are they at the same level as the cardiologists? Came cardiologists eventually interested in these heart rate monitors by the many views on YouTube?

And by the many reviews of DC Rainmaker on his wordpress site, as it has happend to the MRIS? But did the cardiologists not find anything which is worth testing on a regular basis which is reproducable? And is this the reason that DC Rainmaker and Rob ter Horst never deliver numerical and statistical data so that we cannot falsify or veryfy those? Since they do not want to publish data which cannot be reproduced?

We repeat what we have said before: if these heart rate monitors would by highly accurate, would they not be used also by other scientists than sports scientists? But these devices (Apple watches, Polar H10 chest straps, etc.) are not sold by Adinstruments, Biopac, iWorx and g.tec medical. If these devices would be highly accurate and would deliver reproducable data, would these sellers not offer these heart rate monitors as well? But they don’t.

Do sport heart rate monitor companies know these facts since a long time? Do they know since a long time, that the heart rate monitors they sell do not delivers reproducable data which can be published in high quality scientific journals? And as a consequence, they do not allow that we are able to do our own calculations with those? And is this the reason, we are only allowed to look at numbers in Gramin Connect, but we are not allowed to write Python scripts to analyse those data? And is this the reason, we are only allowed to look at Rob ter Horsts accuracy plots on YouTube, but we do not get access to his data in a table with numerical and statistical data? With the consequence, that a Rob ter Horst 2. cannot falsify or verify his data, what eventually would lower the sales rates?

Let us know in the comments below, when you find accuracy studies of consumer grade heart rate monitors which are reproduced by other researchers.