How do I know if a participant is wearing their smartwatch?

How can I tell, by looking at the data, that a participant is actually using their smartwatch or another wearable?

Tags mobile-data data-cleanup participant-compliance umichmetric
Asked by Gabriel Mongefranco on Wed 11/15/23 12:12 PM Last edited Tue 11/28/23 1:39 PM
Sign In to leave feedback or contribute an answer

Answers (2)

Reema Kadri Fri 2/9/24 10:30 AM

With access to the developer API, you may be able to get this in a few ways.  There may be a flag for whether the accelerometer has detected movement during an epoch, for example.  In the many olden days, we had Omrons with a flag in the output where we could see with a Y/N if the device had been moved during a 1 hour that period, 24 hours per day.  Now, we'd then laugh that if you put the device on top of your refrigerator, every time the compressor kicked on, it could trigger that flag, so it couldn't be absolutely precise.  But, anyway, I'd suggest you paw through the API toolkit to see if there's a straight-up flag option, or if you can just piece it together via accelerometer outputs.

No feedback
I wonder if we could standardize this somehow - come up with a formula that works across Fitbit / Apple Health / Garmin APIs to decide whether a person is using their wearable or not. - Gabriel Mongefranco Mon 2/12/24 9:20 AM

Gabriel Mongefranco Wed 11/15/23 12:16 PM

This question came up at the 2023 MeTRIC Symposium. One of the panelists (cannot remember who) mentioned this approach: determine what is an expected wear time, and compare the expected wear time to actual wear time based on the number of minutes or hours that have data in a 24-hour period.

For example, if you know participants may take their watch off for 5-6 hours each day to charge, then your expected wear time could be 18 hours from midnight to midnight. If a participant is taking off their watch for 10 hours, so 14 out of 24, then you know they are below the expected wear time and can mark that day as non-compliant.

Any other approaches??

No feedback