Wrist devices that log heart rate, steps and sleep. Good for resting heart rate and movement; rough for calories, VO₂ max and sleep stages.
Wrist trackers and smartwatches use motion sensors and optical heart-rate sensors (light shone into the skin to detect each pulse) to estimate steps, heart rate and sleep, and on some models blood oxygen, skin temperature, fitness (VO₂ max) and heart rhythm. A few features, such as an ECG app or irregular-rhythm notifications, may be cleared by regulators; most of the rest are wellness estimates never reviewed for medical accuracy.
The value is in trends against your own baseline: resting heart rate, daily movement, and when you sleep.
An umbrella review of 39 meta-analyses (about 164,000 people) found activity trackers raised physical activity by roughly 1,800 steps a day, with small effects on weight and fitness. In a study of 419,297 smartwatch users, among those who got an irregular-pulse alert and then wore an ECG patch, 34% had atrial fibrillation confirmed. Accuracy depends on the metric: heart rate is good at rest and worse during movement, no tested device estimated calories within 20%, and VO₂ max estimates can be off by about ±10 mL/kg/min for an individual.
Pooling 39 systematic reviews covering 163,992 people, activity trackers increased physical activity by about 1,800 steps a day and reduced body weight by about 1 kg; effects on blood pressure, cholesterol and HbA1c were small and often not significant.
Of 419,297 smartwatch wearers, 0.52% received an irregular-pulse notification; among those who then wore an ECG patch (on average 13 days later), 34% had atrial fibrillation, and 84% of alerts during patch wear coincided with it. The study was funded by the device maker.
In 60 volunteers tested against ECG and indirect calorimetry, six of seven wrist devices measured heart rate within 5% during cycling, but none estimated energy expenditure within 20%. Errors were larger during walking and with higher BMI and darker skin.
Across 14 validation studies, exercise-based watch estimates of VO₂ max were unbiased on average but individual errors spanned about ±10 mL/kg/min; estimates from resting data overestimated by about 2.2 mL/kg/min with wider error.