I'm working on developing Cardiogram, the Apple Watch app from which this data is derived, and working with UCSF cardiology doing machine learning on heart rate data.
I was motivated to download your app based on your article so clearly you did a good job explaining the idea :-) I have to say that the Apple Watch has had a profound impact on my health. For the past 7 years, I work out at a gym almost every morning. However I did not realize that I was excessively sedentary for the rest of the day until I started wearing the watch. I now force myself to move around a lot more and sit less.
Every winter, I tend to gain some wait - a few pounds usually. This winter, I have forced myself to go out for walks in the afternoon, no matter how cold it is to get my overall activity up to my target. No weight gain at all. I suppose this could have been achieved with other fitness trackers but the Apple Watch has a very nice gentle way of reminding me about reaching my movement goals by tapping me on my wrist.
The only explanations I can think of are either that Apple Watch owners are a non-representative sample (possibly true. But _that_ non-representative?) or that you do not manage to measure resting heart rate that well (how good is the Apple Watch at measuring heart rate?)
Apple Watch is pretty great at measuring heart rate in my experience. Part of the difference may be the circumstances in which the measurement is taken—in the clinic, "resting heart rate" is typically measured when the patient is sitting, in a controlled setting.
For an app like Cardiogram, your heart rate is being measured throughout your day, "in the wild," every 10 minutes. We do filter out any measurements taken while you were moving (walking, during a workout) and for a few minutes after the movement stops.
But, of course, your heart rate might be higher in real life than it is at the doctor's office. Maybe you've just had coffee. Or you're stressed out driving in traffic. For example, here's mine in bay area traffic: https://twitter.com/bballinger/status/695704441626886145
In Cardiogram itself, we show the user their whole distribution of resting heart rates. What's reported in the blog post is the median of all those measurements.
By the way, I don't think it's known what the "right" answer is here. We do know stress can trigger a heart attack, and coffee can trigger arrhythmias like atrial fibrillation. So perhaps there's a lot of information lurking in the shape of your personal distribution of heart rates. But before now, nobody has really had the data to answer these types of questions.
So you're not actually measuring resting heart rate, just average non-active heartrate. Those sounds like two different things.
My FitBit HR (also using an optical sensor) says my resting HR has varied between about 52 and 59bpm over the past 5 weeks, which passes the sniff test. I tend to be in the high 70s/low 80s when standing and moving about, but when I sit down (to work or relax or whatever), it drops into the low 50s. One that really sticks in my mind was the day I did a free go karting session courtesy of blood donation. Wisely, they had us race first, then donate. I got out of the kart, walked to the waiting area, sat down, and 5 minutes later was called back and given a resting HR of 42.
Resting HR is tricky - I wouldn't consider myself a gym rat but I do run a few times per week and I ride my bike for fitness so I spend far longer than 1 hr above 150 per week.
My resting HR is low 40's. I might hit 65 in traffic....
This analysis is a first step, but before providing personalized recommendations, I think we'd want to do a lot more validation. There are a lot of corner cases in the real world, and particularly in the wake of what's been happening with Theranos and Zenefits, I think it's on startups to really do things right in healthcare.
Is there any research around sum of heart beats as an indicator of life/health? I'm trying to understand if the marathoner's low resting BPM is fewer total beats over time in spite of having temporary higher BPM for exercise?
Vs someone who maybe has resting heart rate of 70, but never exercises and thus doesnt have any high period either.
I'm working on developing Cardiogram, the Apple Watch app from which this data is derived, and working with UCSF cardiology doing machine learning on heart rate data.
Feel free to ask any questions here!