Tracking My Sleep And Resting Heart Rate

So first what did I do? I tracked my sleep for more than six years and I also tracked my resting heart rate every day for the past four years and I’ll talk about this journey of tracking these aspects.

So how did I do it? I’ve been using a number of different self-tracking devices and apps to measure sleep and heart rate. And in this talk I only highlight the results that I acquired from the original classic Fitbit, Basis Smartwatch, the Fitbit Place and the Ōura ring.

But to give you a bit of context, I started sleep tracking with the first generation Fitbit that you see here in 2011, and this means every night putting the Fitbit device into a sleeve that you wear around your wrist and you manually put the device into sleep mode. And here is like the last one and a half years of data of my sleep showing. Simple stuff like sleep duration, sleep onset, and when I wake up and you can observe the variations and changes over time. At the bottom you see a histogram and the distribution of my sleep duration.

And here I compare the three devices that I talked about with my manual tracking for the past one and a half years. The blue line is my manual tracking all sleep, and using a spreadsheet. And then the red line is the Basis watch which is very close to that. The orange line is Ōura, which is giving slightly shorter sleep time. The green is Fitbit Place, which also tends to report a slightly shorter sleep time but the measurements are highly correlated and the devices are doing a pretty decent job at detecting sleep duration.

However, I want to make a brief comment about sleep state detection because all these devices also report that. And as an example, you can see these measurements of generations of REM sleep, and duration of deep sleep made with the Fitbit Place and the Ōura ring. And if they were in agreement they would be on the straight dotted lines that you see. But as you can see on the plot, Ōura tends to report longer REM sleep than Fitbit Place. On the other hand, Fitbit Place tends to report longer deep sleep than the Ōura ring.

So what to make of this? Basically I don’t know which device to trust if any of them when it comes to sleep state detection. My learning is basically that I’m very skeptical about sleep state detection with the present wearables. The measurements are all over the place as you can see.

So that was about sleep, and sleep got me into being interested in the resting heart rate, because the Bases Smartwatch would be on, activity and sleep would also report the daily measurements of resting heart rate. But initially I thought why would I bother, why should I care about that. My resting heart rate is going to be the same all the time, so I expected it would look something like this, and the variations could be due to the inaccuracies of measuring with the optical sensor on the wrist. But the resting heart rate is an indicator of your heart health, so it makes sense and the measurements are important and relevant. But, then on the other hand why measure it on a daily basis, why would the device give it to me on a daily basis. And you even see doctors that claim resting heart rate tends to be stable from day to day.

But what does it look like? This is plotting the first year that I did, and kind of confirmed that it would be stable from day to day. You see there are some variations but what does that mean? Maybe that could be inaccuracies in the optical measurements.

But I was in for a huge surprise when I got to April 2014, as you can see here. I suddenly got this huge and consistent spike in my resting heart rate values. And some of the measurements were actually close to 90, so way way beyond my usual resting heart rate. And that explains why I got this really nasty virus infection that completely knocked me to the ground for weeks. And it took me actually almost eight weeks to fully recover from this, which is actually also reflected in the resting heart rate data, and even the time it took for me to recover was a surprise.

So here is four years of resting heart rate data measurements, and the huge spike in April 2014 still stands out. But you can see other spikes related to illness, but not as dramatic as the one in 2014.

And earlier this year I learned my sort of anecdotal findings about this, the relationship to illness was confirmed in a study performed by the Snyder lab at Stanford. And they reported a lot of things that wearable sensors can actually predict when you are getting sick.

So so far it only shows you data from resting heart rate from one sensor, and it could leave you thinking, so what about the accuracy, what about precision, what about reliability? Can we really trust these measurements?

So I thought some comparisons, and here early last year I got the Ōura ring that I have here and I also got the Fitbit Place smartwatch. And they are similar to Basis because they check your sleep, measure heart rate and report resting heart rate, and tracking with them all allowed me to do some comparisons.

And here you can see like the last one I have here of resting heart rate data with Basis in red and Ōura in yellow. And you can see, generally Basis reports slightly higher values that is compared to Ōura, but they are highly correlated. Fitbit Place in green, have values that are similar to Basis, so they also correlate.

On the top right, there’s a plot where resting heart rate is plotted against Basis resting heart rate.

So data from two independent devices are highly correlated but not the same.

So how did they measure the resting heart rate that is typically a black box, they don’t really tell you. It’s commercial products, but I do know that Basis use the method of measuring resting heart rate 10 minutes before you wake up, and Ōura seems to me to report the lowest value that it gets during the night. Ōura ring has recently enabled the temperature measurements, so it also does that and you get a daily body temperature deviation. And here is an example that Ōura has correctly measured the deviation of plus two and a half degree in Celsius, so pretty high fever which you also see here.

Here’s a comparison with the temperature measurements from Ōura ring on the top aligned with the data on resting heart rate at the bottom. And in particular, notice the event just before New Year’s eve, 2016, the red circle. Again, I had a huge sudden spike in resting heart rate, and when I looked at the temperature data from Ōura, I found a similar spike in body temperature. So this actually shows how I spent my Christmas evening in bed, because I had a 40 degree Celsius fever.

Right, so what have I learned? That is establishing this sort of continuous feedback loop on resting heart rate over four years has made me highly aware of the many different aspects that influences your resting heart rate. And that includes illness like I showed, but also things like stress. If you eat late in the evening it will elevate your metabolism, which will also elevate your resting heart rate. Physical activity plays a role, as does consumption of alcohol if you’re not fully recovered.

And over the years I’ve sort of developed this mental model of resting heart rate values, meaning that I have an awareness at a level now of resting heart rate. I can pretty much predict if my resting heart rate is going to be elevated. And if I cannot predict it it is usually because there is something that I need to be aware of like the onset of illness or stress.

So I hope this talk has not elevated your resting heart rate level, or has put you to sleep. And with those words thank you for your attention.