The task of measuring events, such as heartbeats, is relatively straightforward. When you try to infer information from a combination of measurements, however, it gets a bit trickier. It is possible that you end up missing key factors that are required in order to give an accurate result. And as we’ve mentioned before, we are well along in the process of transitioning from wearables that simply work, to devices that can be relied on to provide information that is accurate enough to use to make significant health and medical decisions.

This transition was part of the motivation behind a new study by the Stanford University School of Medicine. Researchers undertook a test to see how accurate popular fitness bands are. They assembled a team of 60 volunteers to evaluate seven devices that are intended to measure heart rate: Apple Watch, Basis Peak, Fitbit Surge, Microsoft Band, Mio Alpha 2, PulseOn and the Samsung Gear S2. They compared the results from various activities against those measured by “gold standard” clinical devices. The test subjects represented a range of factors: age, height, weight, skin tone, and fitness level. They looked at the results for heart range and energy expended (calories burned) for the different bands. They published their results in The Journal of Personalized Medicine.

The results showed that all but one device (the Samsung Gear S2) were accurate within 5% of the clinical data for heart rate. On the other hand, none of the devices were within the target 20% accuracy for energy expenditure. This is significant, as it means that fewer than 1,600 to more than 2,400 calories burned could be reported as 2,000 calories. That’s too wide a spread to be of much use. And some devices were off by 80% or more. The study also found that the error tended to be greater for male subjects, for subjects with higher body index (BMI) scores or darker skin tone, and for walking.

The bottom line is that you should take the inferred data from your fitness tracker with a large grain of salt. It can be encouraging (or motivating) to see a number for how much energy you’re using throughout the day, but know that these are rough estimates at best. As the wearables manufacturers continue to place more emphasis on accuracy and improve their algorithms that calculate some of these derived measures, we can hope for more trustworthy results in the future.