Last Updated on July 11, 2025 by pg@petergamma.org
Smartwatch sellers made a big effort to make us believe, that their devices are highly accurate. „Very accurate, if not the most accurate monitor“, we could read from DC Rainmaker and 5krunner about the Polar OH1 heart rate monitor. But how accurate is the Polar OH1 in reality?
Smartwatch sellers went even further and engaged the post-doctoral scientist Rob ter Horst to compare smartwatch accuracy with the Polar H10. Rob ter Horst says, the Polar H10 is the „gold standard in science“:
But this is only true for activities which are faster than around 15 km/h. And for us personally, only runs up to around 15 km/h are of interest as physiologists. Since at higher speeds, there are issues with motion artifacts. And according to the g.tec medical support, they do not know of a manufacturer who has resolved the issues with ECG motion artifacts. They can only be reduced.
Smartwatch sellers currently only have one scientist to test the accuracy of smartwatches, and that is Rob ter Horst.
If there would be two, would they find out that smartwatches deliver heart rate data which cannot be reproduced?
Yes, Rob ter Horst says on his channel, that other people tested the smartwatch accuracy next to him. But these people are not independent of Rob ter Horst. And is the fact that only view continue to write papers about smartwatch accuracy not proof enough, that Rob ter Horsts smartwatch accuracy data are not reproduced by other independent scientists? And would we not have such independent scientific studies, if Rob ter Horsts scientific tests would be reproducible?
We wanted to have a closer look at this topic, and we have written a review about it:
What stroke us the most was the choice of the testing protocol of the different papers about smartwatch accuracy:
Every paper chooses a different testing protocol. Is there a stragedy behind this, so that the data of different papers cannot be compared?
Since we have written this review on www.petergamma.org, we can hardly find new papers about smartwatch accuracy.
For scientists much more interesting than smartwatches are ECG devices. For scientific studies, it starts with the TA 220 from iWorx, which is a 3- lead (or channel) ECG device which costs a view thousand dollars.
But since Scott Harden we know that we can have this much cheaper:
Scott uses the sound card of a Windows PC for the analog to digital conversion. But who wants to carry around a Windows PC with his ECG device? In principle, a LattePanda single board computer would do the job. But this is still a device which is not very handy.
With the ADS1115 16-Bit Analog-to-Digital Converter from Texas Instruments:
we have four channels wit 16 bit. And this is all we need for a 3 channel gold standard ECG device. But if we ant to reach the highest quality level, it is eventually better to choose four separate ADS1115 chips. Adafruit sells this chip currently for 15 USD. And on Aliexpress we can find cloned versions of it.
We furthermore need a single board computer for our device, as for instance the Raspberry Pi 3. This architecture has been used in the following paper:
REMOTE WEB BASED ECG MONITORING USING MQTT PROTOCOL
FOR IOT IN HEALTHCARE
For a 3 channel device, we just need 3 AD8232 ECG chips, one ADS1115 chip, and a Raspberry Pi. Or even better, we build 3 completely separated one channel ECG devices, each on a different single board computer, for instance on a Raspberry Pi zero 2 w.
And as a software to get our heart rate, we can use a Python toolbox such as HeartPi, which works also for ECG signals:
https://python-heart-rate-analysis-toolkit.readthedocs.io/en/latest
Now we have a 3 channel ECG device which can compete with the TA 220 from iWorx.
But where this device cannot compete, is with the list of citations of the iWorx TA 220, which is listed here:
But the advantage is, that what we describe here is an open source device. But without scientists who use it, nobody has tested it yet.
What we write here is also only a brain storm from what we have learned in recent years on YouTube. And we made a first search on google scholar about the AD8232 ECG chip, without going to a university library, and look at the papers which are not open access .
We think it is worth to have a deeper look at this topic, and start studying all the papers about the AD8232 ECG chip, if we want to build a highly accurate heart rate monitor device which can be used for scientific studies.
Sparkfun currently sells the AD8232 for around 25 USD. And on Aliexpress we can find a version for 5 USD. With these chips, the costs for multi-channel ECG devices such as the iWorx TA 220 can be strongly reduced.