You might want to think twice before you put a lot of stock in the latest stress charts from your fitness wearable. A recent study from the Netherlands’ Leiden University, published in the Journal of Psychopathology and Clinical Science, has found that when smartwatches and similar devices record readings on stress, fatigue or sleep, they’re frequently getting it wrong.
Researchers studied 800 young adults using the same Garmin Vivosmart 4 smartwatch model. They compared the data the smartwatches produced with the reports that the users created four times per day about how sleepy or stressed they were feeling. Lead author and associate professor Eiko Fried said the correlation between the wearable data and the user-created data was, «basically zero.»
A representative for Garmin did not immediately respond to a request for comment.
Stressed or sex? Your watch doesn’t know
So why do wearables like fitness smartwatches get it so wrong? Their sensors are fairly limited in what they can do. Watches like these need to be worn correctly at all times (a loose or tight watch may give poor readings, for example), and they typically use basic information like pulse rate and movement to make guesses about health.
Those guesses don’t always reflect real-world scenarios. A wearable may identify high stress when the real cause of the change was a workout, excitement over good news, or sex. There are so many potential alternatives to stress or fatigue that the watches in the study never really got it right — and the devices sometimes guessed the complete opposite emotional state from what users recorded.
The Dutch study did note that Garmin’s Body Battery readings, which specifically measure physical fatigue, were more reliable than stress indicators, but still inaccurate. And sleep sensing performed the best of them all, with Garmin watches showing a two-thirds chance of noting the differences between a good night’s sleep and a bad one.
It’s also worth noting that smartwatch sensors can become more accurate as technology improves. It would be interesting to run a similar study with the newer Garmin Vivosmart 5 to see if anything has improved, as well as see if other models like the latest versions of the Apple Watch have similar accuracy results.
