Last Updated on August 3, 2023 by pg@petergamma.org
Rob ter Horst uses his Polar H10 ECG chest strap as a reference device for his accuracy studies:
Reference paper:
https://pubmed.ncbi.nlm.nih.gov/31004219/
The authors if this paper looked at the ECG data after the activity and calculated the signal quality from the ECG signal. They found a signal quality of the Polar H10 chest strap of 99.6 %
If we look at Robs accuracy plots and listen to his arguments, he suggests that the Polar H10 chest strap is the most accurate device followed by Apple watches. But Rob has no statistical and numerical data to offer to proof his statements.
But if we look closely to the lower range of activities, for instance for a run on a treadmill up to 15 km/h, as it was tested in a paper written by cardiologists from the Cleveland Clinic in Ohio:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6732081/
In this study, a cardiologist stood next to the treadmill and checked every ECG peak by eye. Was this done, to control the ECG signal to get a setup which is as optimal as possible?
The result of the paper is:
«Overall, the Polar H7 Chest Strap had the highest agreement with the ECG (rc=98). This was followed by the Apple Watch III (rc=96). «
We can see in this paper, that the Polar H7 had the highest agreement with a 3 lead ECG device which is also used for clinical applications, followed by the Apple Watch III. We do not know of any new data from cardiologists which use a 3 lead ECG reference device which would confirm that for instance on a treadmill up to 15 km/h, the Polar H10 is more accurate than the Polar H7, or newer Apple watches are more accurate than the Apple watch 3.
Is this technically possible to increase the accuracy of chest straps and optical heart rate monitors further than that of the devices which where already tested in these papers? The clinical 3 lead ECG device used in the paper above of the Cleveland Clinic in Ohio is not in doubt as far as accuracy is concerned and defined as a 3 lead gold standard device for HR measurments. But a chest strap as for instance Polar chest straps have a reduced architecture as compared to the clinical and research grade 3 lead ECG devices. And optical sports watches as the Apple watches measure PPG signals. These signals have rounded peaks, the ECG signal have sharp peaks. Do sharp peaks not make it easier to measure accurate data? And also, is it technically possible to increase further the accuracy of optical heart rate monitors and chest straps, or did they already reach the accuracy limit of what is technically possible with these low-cost devices.
Therefore, we are asking the question: Are Rob ter Horsts accuracy plots misleading? If we would look with with a magnifying glass at the data in the lower range of activites, as for instance for runs on a treadmill up to 15 km/h, we know of no new data of cardiologists which use their clinical grade 3 lead ECG device as a reference which would show us that for instance the Polar H10 and the Apple watch Ultra had better accuracy values than the Polar H7 or the Apple watch 3.
The device we introduced here:
includes an OpenBCI module which has the specifications of clinical or research grade ECG device and can eventually be used used for these applications. A paper showed that a 3 lead ECG device based on OpenBCI can be used instead of an Adinstruments 3 lead ECG device for a specific application:
This device could also be helpful to answer the open questions we mentioned here.
One problem which is not discussed in the above papers, but which we have discussed in our journal before are ECG motion artefacts. We suggested a solution how to solve this problem:
By studying the open questions mentioned here and finding answers to those, we could increase the quality of studies about accuracy of consumer grade heart rate monitors, so that these are accepted by sports scientists, cardiologists, physiologists, medical doctors, engineers, and all other heart rate monitor users.