Wearables missed weight training calories even after personal calibration
In short
In 31 trained women, a heart rate plus motion research wearable (Actiheart 5) and an accelerometer (GENEActiv) were compared with indirect calorimetry. With individual calibration, Actiheart matched group-level energy expenditure in intermittent high-intensity running with a mean bias of 4.5 kcal, yet individuals were off by -114 to +123 kcal. For weight training, neither device met the ±10% equivalence bound, and the authors suggested MET-based estimates are more practical.
Plenty of lifters plug their watch's workout calories straight into their diet math. In this study, research wearables were individually calibrated and still failed to estimate weight training energy expenditure reliably.
How was it tested?
- Participants: 31 trained women.
- Devices: Actiheart 5, which combines heart rate and motion, and the wrist-worn GENEActiv accelerometer, each with generic and individually calibrated algorithms.
- Calibration: resting metabolic rate, body measurements and a graded treadmill test for each person.
- Exercise: a modified rugby shuttle test for intermittent high-intensity work, then a resistance training session.
- Reference: indirect calorimetry, plus an anaerobic energy estimate from blood lactate for the lifting session. Within ±10% of the reference counted as equivalent.
Did they get interval running right?
Calibrated Actiheart was equivalent to the reference, with a mean bias of 4.5 kcal (SD 60.4 kcal). Calibrated GENEActiv overestimated by 69.5 kcal on average. Even Actiheart missed individuals by -113.9 to +122.8 kcal, with an intraclass correlation of only 0.62. The group average can be right while your own number is off by more than 100 kcal.
How should you count calories on lifting days?
For resistance training, neither device was valid with either algorithm. Long rest periods and short bursts of effort make heart rate and motion poor proxies for actual energy use. The authors concluded that metabolic equivalent (MET) tables are a more practical estimate. Ultimately, a few weeks of bodyweight trend is the reliable check on intake. Logging bodyweight regularly in fairlift also keeps your Muscle Index calculated on a current weight. What wearables do and do not measure well is also covered in our smart ring piece.
The study covered two research devices and trained women only. It did not test consumer smartwatches directly, but calibrated research devices also missed on weight training.
Frequently asked questions
Do smartwatches measure weight training calories accurately?
In a study of 31 trained women, two heart rate and motion based research wearables both failed to estimate resistance training energy expenditure within ±10% of the reference, even with individual calibration.
Does personal calibration make wearable calories accurate?
For intermittent high-intensity running, calibrated Actiheart passed with a mean bias of 4.5 kcal. Individual errors still ranged from -114 to +123 kcal, and calibration did not fix weight training estimates.
How should I estimate calories burned lifting weights?
The authors concluded MET-based estimates are more practical. Whether your intake is right is best confirmed by your bodyweight trend over several weeks.
Which devices were tested?
Actiheart 5, which combines heart rate and motion, and the wrist-worn GENEActiv accelerometer. Both are research devices.
Source: PubMed