Counts describe this library, not the strength of evidence.
Cómo interpretar la evidencia
Validation is specific to a metric, model, reference and setting. A useful heart-rate measurement does not validate calorie estimates or sleep stages. Even processing the same accelerometer data differently can change estimated activity–disease relationships.
In a national survey of 4551 US adults, about 30% (1266/4551) used wearable health devices, and use was linked to younger age, higher household income, exercise enjoyment and technology self-efficacy. The data are self-reported cross-sectional survey answers, so they describe who uses wearables, not why, and several estimates have wide confidence intervals.
In 100 participants, a commercial smartwatch placed at different body positions agreed closely with a standard 12-lead ECG, with Cohen kappa 0.88 for ST elevation and 0.85 for non-ST elevation changes; sensitivity was 93% and 94% and specificity 95% and 92%. Only 81 people had a heart attack and 19 were healthy, and recordings were taken under supervised conditions, so this is not a test for use at home.
Using smartphone and wearable data from 1062 participants in five European countries, the authors detected lockdown-period changes in time at home, distance travelled, phone use, heart rate and sleep (most P<.001). The sample was a recruited research group, so the patterns may not represent each country's population.
This review of 158 publications on nine commercial wearable brands found that Fitbit, Apple Watch and Samsung appeared to measure steps accurately in laboratory settings, heart rate accuracy varied by brand, and no brand was accurate for energy expenditure. Most evidence was lab-based and concentrated on Fitbit, and device models change quickly.
In 59 healthy children aged 9-11, the Fitbit Charge HR underestimated steps (~11.8 per minute) and heart rate (~3.58 bpm) and identified sleep well (95.8% sensitivity) but wake poorly (56.3% specificity). Sleep was checked in only 26 children, so the sleep accuracy figures rest on a small group.
Daily steps from over 740,000 Withings tracker users fell by 25% to 54% in regions under strict total home confinement, while partial lockdowns showed no clear drop. The data come from tracker owners only and are aggregate, so they show population trends in steps rather than who obeyed the rules.
In a feasibility trial of 30 people with COPD, 47% (14/30) withdrew, often because the technology was hard to use, though those who stayed were mostly positive. The trial was built to test whether a bigger trial is workable, not to prove benefit, and the people who finished were healthier and more used to digital tools.
Across 82 articles in healthy adults, inertial sensors measured step and stride times with excellent validity and reliability, while variability and symmetry measures were poor to moderate. The evidence is thin: the median study had only 12 participants.
Pooling 35 randomised trials, programs built around consumer activity trackers raised physical activity by 2123 steps per day (95% CI 1605-2641) and modestly lowered systolic blood pressure, waist circumference and LDL cholesterol. The programs mostly included counselling, and the abstract does not say how much the tracker itself contributed.
The international journal of behavioral nutrition and physical activity
A review of 19 clinical studies in adults found that wearable sleep monitoring is feasible, but its agreement with polysomnography (the lab sleep test) was limited. Studies varied widely in size, from 1 to 15,839 people, so the picture is uneven.
In 22 women, nocturnal finger skin temperature from an Oura ring was 0.30 °C (SD 0.12) higher in the luteal phase than the follicular phase, and the best algorithm detected ovulation with 83.3% sensitivity. It is a small pilot study, so the accuracy may not generalize to other users.
Among 96 706 UK Biobank participants aged 40 or older, people with chronic disease did 9% (61 minutes; 95% CI, 57.8-64.8) less moderate activity and 11% (3 minutes; 95% CI, 2.7-3.3) less vigorous activity per week than people without. This is a snapshot, so it cannot tell whether disease reduces activity or low activity contributes to disease.
Across 9 randomised trials with 1,352 participants, people with heart disease who used a wearable tracker plus exercise advice improved cardiorespiratory fitness by a mean 1.65 mL/kg/min more than controls (95% CI 0.64-2.66). The pooled step-count result was not significant until a sensitivity analysis was run, and the review covers few, varied trials.
In 20 people with COPD wearing a finger pulse oximeter for 1 week, blood oxygen saturation (SpO2) varied a lot, and 50% (10/20) switched between desaturator and nondesaturator over the week. With only 20 patients and a single week, the results show that spot checks can miss fluctuations, not how often this happens in COPD generally.
A systematic review that screened 857 articles found only 7 studies (5 of mobile health tools, 2 of wearable activity trackers) of interventions to raise physical activity in healthy children and adolescents, and found no evidence of an effect on activity outcomes. That is an absence of evidence rather than proof of no effect, because the studies were few, varied and often at risk of bias.
In 237 conception-seeking Swiss women wearing the Ava bracelet at night, wrist skin temperature, heart rate and respiratory rate shifted across the cycle (all P<.001), and a machine learning model detected the fertile window with 90% accuracy (95% CI 0.89 to 0.92). The accuracy comes from one device and one group of women trying to conceive, so it does not show how well it works for contraception or irregular cycles.
In 10 focus groups of older adults, long-term tracker users cited social support through collaboration as their main reason to keep going, while former users said curiosity about quantifying activity drove their use rather than a wish to be more active. These are qualitative opinions from a few small groups, not measured effects on activity.
Among 2354 Framingham adults, each extra hour of accelerometer-measured light-intensity activity was linked to higher total brain volume, equivalent to about 1.1 years less brain aging (P = .003). This is a cross-sectional snapshot, so it cannot show that light activity protects the brain rather than healthier brains allowing more movement.