Let me say this again: protocols matter.
I’ve covered in an earlier blog, how not only the accuracy of the sensor but also the consistency and timing of the protocol, are key if we want not only to collect accurate data but also to interpret it in a useful way.
I recently brought up a very simple issue with the Apple Watch not being a reliable device for (automated) resting heart rate measurement, and much to my dismay, the point I was making, i.e. protocols matter, was easily missed.
Here I want to make this as simple as possible, with a couple of examples.
What does it mean that the consistency and timing of the protocol are key?
If our goal is to use your resting physiology as a marker of how you respond to stress, then your resting physiology needs to be measured consistently in terms of the protocol used: it must be measured every day under the same circumstances, and as far as possible from stressors. This is why the morning is an ideal moment, and also why the night can work too.
It makes no sense to measure heart rate during the day at random times (i.e. what the Apple Watch does, or even Fitbit to a certain extent), and then interpret it in the context of the stress response. This is clearly highlighted in the quoted example, here: if a person’s resting heart rate in Health goes from 50 to 70 bpm you’d expect major sickness. Instead, it was just the inconsistency of the protocol: the watch collected data when it thought the person was resting (i.e. not moving), but after exercise, when heart rate was elevated.
Wearables do not have context (sometimes, quite surprisingly, even if you used the same wearable to track your activity), and therefore can make foolish assumptions and make the interpretation of the data faulty.
Someone brought up that by definition, in medicine, resting heart rate is your heart rate measured while awake, and therefore it is a good thing that the Apple Watch excludes night data to provide you with your resting heart rate.
Awake does not mean randomly during the day (can’t believe I had to write this): we use protocols for a reason, and the inconsistency of the protocol is the problem here, as it makes the interpretation meaningless (as shown in the example quoted).
The goal here is not “to measure heart rate while you are not moving”, the goal is to assess your resting physiology, and to do that, we need to measure at rest, far from stressors, in repeatable conditions every day: i.e., according to a certain protocol.
Let’s make this even simpler, moving away from resting heart rate or HRV and using parameters that are more widely measured and used in medicine.
More examples: blood pressure and weight
If your doctor tells you to measure your blood pressure, ideally you'd measure it first thing in the morning, at rest, while awake, and before doing anything that can impact the data (having breakfast, coffee, exercising, etc.).
You would not measure your blood pressure after exercise when you are not moving (i.e. the engineer's definition of resting, through an accelerometer, often how wearables end up working), and report it to your doctor as your resting blood pressure. You would not measure it after walking up five flights of stairs with your groceries and report it to your doctor as your resting blood pressure. You would not measure it after a hot bath, and report it to your doctor as your resting blood pressure. Get it? This is what wearables do when they measure randomly, without a protocol.
The same with body weight: you would not measure your body weight after lunch, or after sweating 3 hours on a ride, and consider it your body weight for that day. You would measure it after waking up, in standardized conditions each day (before eating, and exercising for example).
Protocols matter.
It is the same for resting heart rate: if you want to use the Apple Watch to measure your resting heart rate, intentionally, first thing in the morning, that is great and the device works well.
If you use automatically collected data from the watch, that is labeled as resting heart rate, then that's often meaningless, for the reasons discussed here: there is no consistency in the protocol, and protocols matter if we want to interpret the data as opposed to just collecting it.
I hope this helps make use of the data you collect and pick the right wearable for the job.
Fanboys will be fanboys
Personally, I cannot care less about what wearable you decide to use, or if you like the Apple Watch or not, so please, do not take this too hard (I do use an iPhone and have many Apple devices that I love, but I can still think critically about issues with a specific product for a specific application, go figure).
Make sure to use the device for what it can do, as there are always limitations, no matter how pretty or heavily marketed a device is, and no matter how well it can do other things.
Take it easy!
See also:
Marco holds a PhD cum laude in applied machine learning, a M.Sc. cum laude in computer science engineering, and a M.Sc. cum laude in human movement sciences and high-performance coaching.
He has published more than 50 papers and patents at the intersection between physiology, health, technology, and human performance.
He is co-founder of HRV4Training, advisor at Oura, guest lecturer at VU Amsterdam, and editor for IEEE Pervasive Computing Magazine. He loves running.
Social:
Twitter: @altini_marco.
Personal Substack.