One of the key components of WHOOP is the ability to quantify the strain your body takes on. Our Strain metric allows you to understand the cardiovascular load of everything you do over the course of an entire day. It also lets you compare the effect different activities have on your body–how does your daily morning swim stack up against your after-work CrossFit session? Or what about your Saturday round of golf? WHOOP Strain gives you a number for all of these.
Previously, if you wanted to see the Strain of any given activity, you had to manually enter in the app when that activity began and ended. Now, with a new feature released this week, WHOOP can automatically detect activities for you.
Here’s How it Works
If WHOOP notices a period of elevated heart rate and movement lasting at least 15 minutes and earning an 8.0 or higher on our 0-21 Strain scale, the text “Analyzing Elevated HR” will appear on the “Overview” screen of your app:
The above video details how this feature can be used on the iPhone app, but it works the same way with Android as well. Regardless, your activity will still be detected and logged even if you don’t check the app immediately afterwards. However, WHOOP will wait until your heart rate has returned to a normal level for a significant period of time before processing the activity on its own (to ensure that you’re finished and not simply taking a break). In this case, it may take up to 20 minutes from the time the activity is completed for it to be logged.
If you’d prefer not to have activities logged automatically, you can go to the “Settings” page in the app and turn off “Activity Detection.”
How Does WHOOP Do This?
As we went about creating this feature, one of the most difficult aspects of automating activity reporting was determining what should qualify. For example, if you walk down the street at a brisk pace for 20 minutes, does that count? From a biometric standpoint, it does–you’ve got an elevated heart rate with a good amount of associated motion. But from a semantic standpoint, not many of our athlete users would consider that a “workout.”
Unlike with sleep auto-detection, “activity” is not a universally-recognized physiological state that we can use objective data to identify–it’s a mindset which our research has shown is hard to get a consensus of. Activities and workouts are all subjective to each user’s definition. Take a look at the poll below, which activity are you most likely to consider a workout?
Assuming you did all of these things, which would you be most likely to log as an activity on WHOOP?
— WHOOP (@whoop) October 24, 2017
So, what did we do to get around this?
We started by examining user data from thousands of hours of logged workouts. That data was then set against an equally large amount of data from periods of activity that our users did not consider workouts. From there, we were able to use machine learning to build a model that could differentiate between the combinations of motion and heart rate data our users thought were activities worthy of logging, and those they did not.
What are the Benefits?
If you’re someone who regularly logs activities in the WHOOP app, this feature obviously makes tracking your workouts much easier. You no longer have to worry about starting and stopping them manually, or remembering the exact times you exercised if you decide to input them after the fact. WHOOP now does all the work for you.
On the other hand, maybe you’ve used different methods for monitoring workouts in the past, or for whatever reason haven’t cared to log activities on WHOOP. If this is the case, activity auto-detection will now offer you a variety of new insights in your efforts to optimize performance.
For one, it will help you compare and contrast the various activities you do over time. Maybe your 45-minute session in the gym yesterday was more strenuous than the same workout was today? Or, maybe the 10-mile run you do four times a week puts more Strain on your body in the heat of the summer than it does the rest of the year?
We expect many WHOOP users will benefit most from activity auto-detection by gaining a better comprehension of the things they do over the course of a day that cause Strain. A common theme we’ve found in this area is that users who bike to work rarely log their commutes as activities. Oftentimes these commutes may be as strenuous as other things they consider “workouts.”
Who knows, maybe mowing your lawn next Saturday will create as much Strain as the pickup basketball game you play in on Sunday? Now you’re about to find out.