What does the research show about wrist heart rate and calorie estimates?

Wrist heart rate and calorie estimates

Pooled validation studies report wrist heart rate within about 1 beat per minute of ECG or a chest strap on average at rest and on a treadmill, with larger errors during cycling and resistance exercise. Calorie estimates did not match indirect calorimetry in pooled reviews, with single-study errors commonly above 20%. Activity trackers raised daily steps in trials.

Checked 2026-11-13

Status: Sold in the US as general wellness products. FDA's January 2026 guidance lists logging or tracking exercise activity and monitoring heart rate during recreational activity as general wellness claims, and states that the agency does not intend to examine low risk general wellness products for compliance with premarket review requirements [1].
Measures: Heart rate, from light shone into the skin of the wrist to detect each pulse, and calories burned, estimated by each maker's own formula from heart rate, movement and body details.
Also called: wrist heart rate monitor, optical heart rate sensor, PPG heart rate, calories burned, active calories, energy expenditure estimate, fitness tracker, smartwatch, Apple Watch, Fitbit, Garmin, Polar, Samsung Galaxy Watch
Used for: heart rate tracking, calories burned, weight loss, training zones, fitness tracking, daily activity, resting heart rate

Strong several reviews agree · Moderate one review or consistent trials · Limited one or a few small studies · Untested no study in people, listed with what it rests on

Matched the reference method2

StrongShown to estimate heart rate within 1 beat per minute of the reference at rest and on a treadmill, reading 5 to 7 beats low during cycling and resistance training.
LimitedOne study of 60 adults found six of seven wrist devices had a median heart rate error below 5% during cycling, with higher error while walking.

Reported risks3

LimitedLinked to follow-up testing: in one review of 264 patients seen after an Apple Watch abnormal pulse reading, 11.4% received a clinically actionable heart diagnosis.
LimitedLinked to more symptom monitoring, treatment concern and health care use in one study of 172 patients with atrial fibrillation who did or did not use a wearable.
LimitedCase reports: allergic contact dermatitis from wearable device materials, with acrylates, methacrylates and colophonium the allergens most often found on patch testing.

How it compares with the reference method

Matched the reference method2

Limited

Heart rate, Apple Watch, Basis Peak, Fitbit Surge, Microsoft Band, Mio Alpha 2, PulseOn and Samsung Gear S2

One study of 60 adults found six of seven wrist devices had a median heart rate error below 5% during cycling, with higher error while walking.

The numbers and sources

Device: Apple Watch, Basis Peak, Fitbit Surge, Microsoft Band, Mio Alpha 2, PulseOn, Samsung Gear S2
Compared with: Continuous ECG telemetry
Effect: Six of the seven devices had a median heart rate error below 5% during cycling. Error was lowest for cycling and highest for walking, and was higher in men, at greater body mass index and with darker skin tone. The authors concluded most of the devices adequately measured heart rate in laboratory activities [5].
Population: 60 volunteers (31 women), mean age 38, chosen to vary in skin tone, body size and fitness [5].
Sources: [5]
Caveats: Published in 2017 with devices from that period. Laboratory sitting, walking, running and cycling only [5].

Strong

Heart rate, wrist-worn optical devices pooled

Shown to estimate heart rate within 1 beat per minute of the reference at rest and on a treadmill, reading 5 to 7 beats low during cycling and resistance training.

The numbers and sources

Device: Wrist-worn optical heart rate devices from 15 brands, pooled [8]. Consumer wearables pooled across 24 systematic reviews [9]. Nine commercial brands across 158 publications [10].
Compared with: ECG or a chest strap
Effect: Across 44 articles and 738 effect sizes, the mean difference from the reference was -0.40 beats per minute during sleep (95% CI -1.64 to 0.83), -0.01 at rest and -0.51 during treadmill walking and running (95% CI -1.60 to 0.58). It was -4.55 during cycling (95% CI -7.24 to -1.87) and -7.26 during resistance training (95% CI -10.46 to -4.07), where the difference grew by 3 beats for every 10 beat rise in heart rate. The authors concluded validity was acceptable for most common activities [8]. An umbrella review of 24 systematic reviews reported a mean bias of plus or minus 3% for heart rate [9]. A systematic review found heart rate accuracy in laboratory settings varied by manufacturer and device type [10].
Population: 44 articles [8]. 24 systematic reviews covering 249 validation studies [9]. 158 publications up to May 2019 [10].
Sources: [8], [9], [10]
Caveats: Average differences near zero do not describe the error of a single reading. The umbrella review estimated that about 11% of consumer wearables released had been validated for at least one measure [9]. One review searched only to May 2019 [10].

Mixed results8

Limited

Heart rate, Apple Watch 6, Fitbit Sense and Polar Vantage V

Mixed results: one study of 60 young adults found heart rate error under 5% for Apple Watch 6 in five activities and up to 11% for the other two watches.

The numbers and sources

Device: Apple Watch 6, Fitbit Sense, Polar Vantage V
Compared with: Polar H10 chest strap, not ECG
Effect: Across sitting, walking, running, resistance exercises and cycling, the coefficient of variation for heart rate was under 5% in all five activities for Apple Watch 6, 2.44 to 8.80% for Polar Vantage V and 4.14 to 10.76% for Fitbit Sense [2].
Population: 60 healthy young adults (30 women), mean age 24.9 [2].
Sources: [2]
Caveats: The reference was a chest strap. Young, lean, healthy volunteers [2].

Moderate

Heart rate, Apple Watch models pooled

Mixed results: a meta-analysis found Apple Watch heart rate within 0.3 beats per minute of the reference on average, with limits of agreement of about 7 beats either way.

The numbers and sources

Device: Apple Watch models pooled across generations
Compared with: The criterion method used in each study
Effect: Bland-Altman meta-analysis gave a mean bias of -0.27 beats per minute (95% CI -0.72 to 0.17) with limits of agreement from -7.19 to 6.64. The authors described this as a small underestimation with moderate measurement variability, and reported that accuracy varied with measurement conditions and individual physiology [3].
Population: 82 studies of 14 Apple Watch health metrics, pooled mean age 41.3 [3].
Sources: [3]
Caveats: Studies span many watch generations and software versions. The review is registered as a living review that its authors plan to update yearly [3].

Limited

Heart rate, Apple Watch Series 4, Fitbit Versa, Garmin Fenix 5 and Polar Vantage V

Mixed results: one study of 25 adults found heart rate error from 1% to 19% across four watches, depending on device and running speed.

The numbers and sources

Device: Apple Watch Series 4, Fitbit Versa, Garmin Fenix 5, Polar Vantage V
Compared with: Polar H7 chest strap, not ECG
Effect: Across sitting, walking, running at four speeds and intermittent sprints, the coefficient of variation for heart rate was 0.9 to 4.3% for Apple Watch Series 4, 2.2 to 6.7% for Polar Vantage V, 2.9 to 9.2% for Garmin Fenix 5 and 4.1 to 19.1% for Fitbit Versa. The authors judged the Apple and Polar watches suitable at the intensities tested and advised caution with the Garmin and Fitbit readings at certain intensities [4].
Population: 25 adults, each wearing all four watches [4].
Sources: [4]
Caveats: The reference was a chest strap. Five minutes per activity in one session. Firmware versions are not reported in the abstract [4].

Moderate

Heart rate, Fitbit models pooled

Mixed results: a meta-analysis found Fitbit devices read 3 beats per minute below the reference on average, with limits of agreement from 24 below to 18 above.

The numbers and sources

Device: Fitbit models that combine a heart rate sensor with a motion sensor, pooled
Compared with: Reference heart rate measures
Effect: Across 74 comparisons, Fitbit devices underestimated heart rate by a mean 2.99 beats per minute, with population limits of agreement from -23.99 to 18.01. The authors concluded this underestimation can be acceptable on average for heart rate, and that results varied with study quality, participant age, activity type and Fitbit model [6].
Population: 52 studies in the review, 41 in the meta-analysis [6].
Sources: [6]
Caveats: Figures are after removing studies at high risk of bias. Models and firmware are pooled [6].

Limited

Heart rate, Polar Vantage V2 and Polar Verity Sense

One study of 16 adults found a 1.4% heart rate error for an optical band on the upper arm and 6.8% for a watch on the wrist.

The numbers and sources

Device: Polar Verity Sense, worn on the upper arm, and Polar Vantage V2, worn on the wrist
Compared with: Polar H10 chest strap
Effect: Across nine activities from lying down to high-intensity intervals, the upper-arm band had a bias of -0.05 beats per minute, a mean absolute error of 1.43 and a mean absolute percentage error of 1.35%. The wrist watch on the non-dominant wrist had a bias of 2.56 beats per minute, a mean absolute error of 6.41 and a mean absolute percentage error of 6.82% (concordance 0.92), and its accuracy varied considerably by activity [7].
Population: 16 healthy adults (7 women), mean age 27.4, two sessions each [7].
Sources: [7]
Caveats: Sixteen participants and a chest strap reference. The authors declared no conflicts of interest [7].

Moderate

Heart rate across skin tones, wrist-worn devices

Mixed results: in one systematic review, four of ten studies found lower heart rate accuracy with darker skin, four found no effect and two were mixed.

The numbers and sources

Device: Wrist-worn consumer devices pooled [11]. Apple Watch 4, Fitbit Charge 2, Garmin Vivosmart 3 and Xiaomi Mi Band 3, with two research devices [12]. Apple Watch Series 9 [13].
Compared with: ECG [12]. A Polar H7 chest strap [13]. Reference methods varied across the review [11].
Effect: Of 10 studies that reported heart rate or rhythm accuracy by race or skin tone, 4 found significantly lower accuracy in darker-skinned participants, 4 found no effect and 2 were mixed [11]. In 53 people spread evenly across six skin tone categories, accuracy did not differ significantly by skin tone, while it differed between devices and absolute error was on average 30% higher during activity than at rest [12]. In 30 young adults in three skin pigmentation groups, Apple Watch Series 9 heart rate varied more from the chest strap in the darkest group [13].
Population: 469 participants in 10 studies [11]. 53 adults aged 18 to 54 [12]. 30 healthy young adults [13].
Sources: [11], [12], [13]
Caveats: The review's authors call the evidence preliminary and inconclusive and note small samples and subjective skin tone grading [11]. One abstract gives no error figures by group [13].

LimitedStudied in men

Heart rate during resistance exercise, Apple Watch Series 8, Fitbit Charge 5, Garmin Vivosmart 5 and Samsung Galaxy Watch 5

Mixed results: one study of 70 men found Apple Watch Series 8 within 2 beats of ECG during lifting, while two watches read 20 to 26 beats low after sets.

The numbers and sources

Device: Apple Watch Series 8, Fitbit Charge 5, Garmin Vivosmart 5, Samsung Galaxy Watch 5
Compared with: ECG
Effect: For mean heart rate across the session, Apple Watch Series 8 had a bias of -0.36 beats per minute, a mean absolute error of 1.81 and an intraclass correlation of 0.969. For dynamic heart rate across the protocol, its mean absolute error was 4.95 beats per minute (intraclass correlation 0.904). Immediately after each set, the Samsung and Fitbit devices underestimated ECG heart rate by 25.60 and 19.59 beats per minute [14].
Population: 70 healthy men aged 20 to 39 performing bench press, T-bar row, squat and deadlift at 10-repetition maximum loads [14].
Sources: [14]
Caveats: Men only. The abstract gives full figures only for the Apple Watch [14].

Limited

Energy expenditure, Apple Watch Series 6, Garmin Fenix 6 and Huawei Watch GT 2e

Mixed results: one study of 20 adults found calorie errors of 10 to 32% for walking and 12 to 24% for running across three watches.

The numbers and sources

Device: Apple Watch Series 6, Garmin Fenix 6, Huawei Watch GT 2e
Compared with: Indirect calorimetry with a COSMED K5 portable system
Effect: For walking 2 km at 6 km/h, mean absolute percentage error was 19.8% for Apple Watch, 32.0% for Garmin and 9.9% for Huawei, with intraclass correlations of 0.821, 0.216 and 0.760. For running 2 km at 10 km/h, error was 24.4%, 21.8% and 11.9%. The Apple and Garmin watches read significantly higher than calorimetry in both tests, and the Huawei watch did not differ. The authors described validity as moderate [15].
Population: 20 young normal-weight Chinese adults on an outdoor 400 meter track [15].
Sources: [15]
Caveats: Twenty participants and two fixed speeds [15].

Did not match the reference method8

Limited

Energy expenditure, Apple Watch 6, Fitbit Sense and Polar Vantage V

One study of 60 young adults found calorie estimates from three watches varied 13 to 30% from indirect calorimetry across five activities.

The numbers and sources

Device: Apple Watch 6, Fitbit Sense, Polar Vantage V
Compared with: Indirect calorimetry with a Metamax 3B portable system
Effect: The coefficient of variation for energy expenditure was 14.68 to 24.85% for Apple Watch 6, 16.54 to 25.78% for Polar Vantage V and 13.44 to 29.66% for Fitbit Sense across sitting, walking, running, resistance exercises and cycling [2].
Population: 60 healthy young adults (30 women), mean age 24.9 [2].
Sources: [2]
Caveats: Young, lean, healthy volunteers in a laboratory [2].

Moderate

Energy expenditure, Apple Watch models pooled

Not shown to match reference energy expenditure. A systematic review found Apple Watch errors from 10% during running to 152% during walking across 8 studies.

The numbers and sources

Device: Apple Watch models pooled. Five of the eight studies tested Series 2 or older.
Compared with: Criterion measures of energy expenditure
Effect: Across 8 studies, mean absolute percentage error ranged from 9.71% during running to 151.66% during walking. All six studies that calculated it reported 20% or higher in at least one test condition. The authors described error for energy expenditure as inconsistent and frequently large, at rest and during exercise [3].
Population: 270 participants in 8 studies, 63% male, mostly young physically active adults [3].
Sources: [3]
Caveats: Most of the evidence comes from early watch generations. No trend in error by exercise intensity was seen [3].

Limited

Energy expenditure, Apple Watch Series 4, Fitbit Versa, Garmin Fenix 5 and Polar Vantage V

One study of 25 adults found calorie estimates from four watches varied 8 to 35% from indirect calorimetry across sitting, walking and running.

The numbers and sources

Device: Apple Watch Series 4, Fitbit Versa, Garmin Fenix 5, Polar Vantage V
Compared with: Indirect calorimetry
Effect: The coefficient of variation for energy expenditure was 13.5 to 27.1% for Apple Watch Series 4, 16.3 to 28.0% for Polar Vantage V, 15.9 to 34.5% for Garmin Fenix 5 and 8.0 to 32.3% for Fitbit Versa, depending on intensity. The authors concluded none of the four watches was suitable for monitoring energy expenditure at the intensities and durations tested [4].
Population: 25 adults, 5 minutes each of sitting, walking, running at four speeds and intermittent sprints [4].
Sources: [4]
Caveats: Five minute bouts in a single session [4].

Limited

Energy expenditure, Apple Watch Series 8, Fitbit Sense 2, Garmin Forerunner 955 and Samsung Galaxy Watch 5

One study of 58 adults found three of four watches overestimated cycling calories by 22 to 69 kcal over a short session, with error rising with body fat.

The numbers and sources

Device: Apple Watch Series 8, Fitbit Sense 2, Garmin Forerunner 955, Samsung Galaxy Watch 5
Compared with: Indirect calorimetry with a COSMED K5 system
Effect: Mean bias was 21.60 kcal for Apple (95% CI 11.59 to 31.60), 68.61 kcal for Garmin (95% CI 53.28 to 83.94), 56.76 kcal for Samsung (95% CI 45.11 to 68.41) and 3.14 kcal for Fitbit (95% CI -8.96 to 15.24). Error increased with body fat percentage across all four brands (p < .01). The authors concluded the watches substantially misestimated activity energy expenditure [16].
Population: 58 Hispanic adults with Fitzpatrick skin types III to V, in a 10 minute recumbent cycling protocol with 5 minutes of rest before and after [16].
Sources: [16]
Caveats: The non-significant Fitbit bias depended on device-specific outlier removal, and usable pairings ranged from 44 (Fitbit) to 52 (Apple). One short cycling session [16].

Limited

Energy expenditure, Apple Watch, Basis Peak, Fitbit Surge, Microsoft Band, Mio Alpha 2, PulseOn and Samsung Gear S2

One study of 60 adults found none of seven wrist devices estimated energy expenditure with error below 20%.

The numbers and sources

Device: Apple Watch, Basis Peak, Fitbit Surge, Microsoft Band, Mio Alpha 2, PulseOn, Samsung Gear S2
Compared with: Indirect calorimetry
Effect: No device achieved an error in energy expenditure below 20% across sitting, walking, running and cycling. Error was higher in men, at greater body mass index, with darker skin tone and while walking [5].
Population: 60 volunteers (31 women), mean age 38 [5].
Sources: [5]
Caveats: Published in 2017 with devices from that period [5].

Moderate

Energy expenditure, Fitbit models pooled

Not shown to match reference energy expenditure. A meta-analysis found Fitbit devices read 2.8 kcal per minute low on average, with wide limits of agreement.

The numbers and sources

Device: Fitbit models that combine a heart rate sensor with a motion sensor, pooled
Compared with: Reference measures of energy expenditure
Effect: Across 29 comparisons, Fitbit devices underestimated energy expenditure by a mean 2.77 kcal per minute, with population limits of agreement from -12.75 to 7.41 kcal per minute. The authors concluded the energy expenditure measure "may be inaccurate for some research purposes" [6].
Population: 52 studies in the review, 41 in the meta-analysis [6].
Sources: [6]
Caveats: Figures are after removing studies at high risk of bias. Models and firmware are pooled [6].

Strong

Energy expenditure, wrist and arm devices pooled

Not shown to match laboratory measures of energy expenditure in pooled reviews, with errors that ranged from 21% under to 15% over and varied by activity.

The numbers and sources

Device: Wrist-worn and arm-worn activity monitors pooled [17]. Consumer wearables pooled across systematic reviews [9]. Nine commercial brands [10].
Compared with: Indirect calorimetry, room calorimeters and doubly labeled water
Effect: A meta-analysis of 60 studies and 104 effect sizes found accuracy varied by activity type, with large heterogeneity for many devices (I² above 75%). Adding heart rate or heat sensing to motion sensing reduced error in most activity types [17]. An umbrella review reported that wearables mostly underestimated energy expenditure, with a mean bias of -3% and error ranging from -21.27% to 14.76% [9]. A systematic review of 158 publications concluded that for energy expenditure "no brand was accurate" [10].
Population: Healthy adults in 60 studies [17]. 24 systematic reviews [9]. 158 publications up to May 2019 [10].
Sources: [17], [9], [10]
Caveats: Each maker uses its own undisclosed formula, and reviews pool device generations. One abstract reports no pooled error figure [17].

LimitedStudied in men

Energy expenditure during resistance exercise, Apple Watch Series 8, Fitbit Charge 5, Garmin Vivosmart 5 and Samsung Galaxy Watch 5

One study of 70 men found calorie estimates from four watches barely tracked indirect calorimetry during a lifting session, with intraclass correlations of 0.02 to 0.03.

The numbers and sources

Device: Apple Watch Series 8, Fitbit Charge 5, Garmin Vivosmart 5, Samsung Galaxy Watch 5
Compared with: Indirect calorimetry
Effect: Intraclass correlation between each watch's workout-session energy expenditure and indirect calorimetry was 0.021 to 0.033 across the four devices [14].
Population: 70 healthy men aged 20 to 39 performing four barbell exercises [14].
Sources: [14]
Caveats: Men only, one session [14].

Does using one change anything

Shown to help2

Strong

Daily steps and physical activity

Shown to raise physical activity by about 1,800 steps a day across 39 systematic reviews of activity trackers in healthy and clinical groups.

The numbers and sources

Effect: An umbrella review of 39 systematic reviews and meta-analyses found activity trackers improved physical activity (SMD 0.3 to 0.6), equal to about 1,800 extra steps a day and 40 minutes a day more walking, and improved fitness (SMD 0.3) [18].
Population: 163,992 participants of all age groups, from healthy and clinical populations [18].
Sources: [18]
Caveats: The reviews cover wearable activity trackers as a class. They do not test whether heart rate or calorie readings themselves changed behavior [18].

Strong

Body weight

Shown to lower body weight by about 1 kg across systematic reviews of activity tracker trials.

The numbers and sources

Effect: The same umbrella review found improvements in body composition (SMD 0.7 to 2.0), equal to reductions of about 1 kg in body weight [18].
Population: Participants in 39 systematic reviews and meta-analyses [18].
Sources: [18]
Caveats: The reviews cover wearable activity trackers as a class and do not isolate wrist heart rate or calorie readings [18].

Mixed results1

Strong

Blood pressure, cholesterol and blood sugar

Mixed results: effects of activity trackers on blood pressure, cholesterol and HbA1c across systematic reviews were typically small and often not statistically significant.

The numbers and sources

Effect: Effects of activity trackers on blood pressure, cholesterol and glycosylated hemoglobin were typically small and often non-significant, as were effects on quality of life and pain [18].
Population: Participants in 39 systematic reviews and meta-analyses [18].
Sources: [18]
Caveats: The abstract gives no pooled figures for these outcomes [18].

Risks

Reported risks3

Limited

Medical testing after abnormal pulse alerts

Linked to follow-up testing: in one review of 264 patients seen after an Apple Watch abnormal pulse reading, 11.4% received a clinically actionable heart diagnosis.

The numbers and sources

Effect: Among 264 patients evaluated over 4 months for an abnormal pulse detected with an Apple Watch, 41 (15.5%) had an explicit alert documented and 58 (22.0%) already had atrial fibrillation. Testing included a 12-lead recording in 59.8%, a Holter monitor in 29.2% and a chest x-ray in 24.2%. A clinically actionable cardiovascular diagnosis was made in 30 patients (11.4%) [19].
Population: 264 patients evaluated over a 4 month period, reviewed retrospectively [19].
Sources: [19]
Caveats: A retrospective chart review with no comparison group, so it cannot show how many tests would have happened without the watch. Published in 2020 [19].

Limited

Symptom preoccupation and health care use in people with atrial fibrillation

Linked to more symptom monitoring, treatment concern and health care use in one study of 172 patients with atrial fibrillation who did or did not use a wearable.

The numbers and sources

Effect: Compared with non-users, wearable users reported higher rates of symptom monitoring and preoccupation (P = 0.03) and more concerns about treatment (P = 0.02). Of wearable users, 20% reported anxiety and always contacting their doctor in response to irregular rhythm notifications. After matching, users had more atrial fibrillation care over 9 months (P = 0.04), including more ECGs, echocardiograms and ablation procedures [20].
Population: 172 patients with atrial fibrillation, mean age 72.6, 42% women, of whom 83 used a wearable [20].
Sources: [20]
Caveats: Observational and retrospective, in older patients with a diagnosed heart rhythm condition. It cannot show the wearable caused the difference, and the authors call for randomized studies [20].

Limited

Skin reactions under the device

Case reports: allergic contact dermatitis from wearable device materials, with acrylates, methacrylates and colophonium the allergens most often found on patch testing.

The numbers and sources

Effect: A review of allergic contact dermatitis from wearable health devices reported that rash from wearable activity trackers is commonly reported by consumers, and that acrylates, methacrylates and colophonium were the allergens most often identified on patch testing, found in adhesives or in the units themselves [21].
Population: Published case reports and series of users with skin reactions [21].
Sources: [21]
Caveats: A narrative review that also covers adhesive glucose sensors. It gives no rate, and its authors suggest reactions may be underreported [21].

Sources

US Food and Drug Administration. General Wellness: Policy for Low Risk Devices. Guidance for Industry and Food and Drug Administration Staff. January 6, 2026. https://www.fda.gov/regulatory-information/search-fda-guidance-documents/general-wellness-policy-low-risk-devices
Hajj-Boutros G, Landry-Duval MA, Comtois AS, et al. Wrist-worn devices for the measurement of heart rate and energy expenditure: A validation study for the Apple Watch 6, Polar Vantage V and Fitbit Sense. European Journal of Sport Science. 2023. https://doi.org/10.1080/17461391.2021.2023656
Lambe R, Baldwin M, O'Grady B, et al. The accuracy of Apple Watch measurements: a living systematic review and meta-analysis. npj Digital Medicine. 2026. https://doi.org/10.1038/s41746-025-02238-1
Düking P, Giessing L, Frenkel MO, et al. Wrist-Worn Wearables for Monitoring Heart Rate and Energy Expenditure While Sitting or Performing Light-to-Vigorous Physical Activity: Validation Study. JMIR mHealth and uHealth. 2020. https://doi.org/10.2196/16716
Shcherbina A, Mattsson CM, Waggott D, et al. Accuracy in Wrist-Worn, Sensor-Based Measurements of Heart Rate and Energy Expenditure in a Diverse Cohort. Journal of Personalized Medicine. 2017. https://doi.org/10.3390/jpm7020003
Chevance G, Golaszewski NM, Tipton E, et al. Accuracy and Precision of Energy Expenditure, Heart Rate, and Steps Measured by Combined-Sensing Fitbits Against Reference Measures: Systematic Review and Meta-analysis. JMIR mHealth and uHealth. 2022. https://doi.org/10.2196/35626
Schweizer T, Gilgen-Ammann R. Wrist-Worn and Arm-Worn Wearables for Monitoring Heart Rate During Sedentary and Light-to-Vigorous Physical Activities: Device Validation Study. JMIR Cardio. 2025. https://doi.org/10.2196/67110
Zhang Y, Weaver RG, Armstrong B, et al. Validity of Wrist-Worn photoplethysmography devices to measure heart rate: A systematic review and meta-analysis. Journal of Sports Sciences. 2020. https://doi.org/10.1080/02640414.2020.1767348
Doherty C, Baldwin M, Keogh A, et al. Keeping Pace with Wearables: A Living Umbrella Review of Systematic Reviews Evaluating the Accuracy of Consumer Wearable Technologies in Health Measurement. Sports Medicine. 2024. https://doi.org/10.1007/s40279-024-02077-2
Fuller D, Colwell E, Low J, et al. Reliability and Validity of Commercially Available Wearable Devices for Measuring Steps, Energy Expenditure, and Heart Rate: Systematic Review. JMIR mHealth and uHealth. 2020. https://doi.org/10.2196/18694
Koerber D, Khan S, Shamsheri T, et al. Accuracy of Heart Rate Measurement with Wrist-Worn Wearable Devices in Various Skin Tones: a Systematic Review. Journal of Racial and Ethnic Health Disparities. 2023. https://doi.org/10.1007/s40615-022-01446-9
Bent B, Goldstein BA, Kibbe WA, Dunn JP. Investigating sources of inaccuracy in wearable optical heart rate sensors. npj Digital Medicine. 2020. https://doi.org/10.1038/s41746-020-0226-6
Chase SE, Liddell RG, McGonagle CL, Ives SJ. Accuracy of the Apple Watch Series 9 for Measures of Energy Expenditure and Heart Rate at Rest and During Exercise: Impact of Skin Pigmentation. Journal of Functional Morphology and Kinesiology. 2024. https://doi.org/10.3390/jfmk9040275
Yun JH, Ko JM, Lee TH. Criterion Validity of Smartwatches for Measuring Heart Rate and Energy Expenditure During Resistance Exercise. Bioengineering. 2026. https://doi.org/10.3390/bioengineering13101099
Le S, Wang X, Zhang T, et al. Validity of three smartwatches in estimating energy expenditure during outdoor walking and running. Frontiers in Physiology. 2022. https://doi.org/10.3389/fphys.2022.995575
Kostrna J, Oparina E, Palacios C, et al. Body fat, skin tone, and the accuracy of smartwatch caloric expenditure estimates. PLOS One. 2026. https://doi.org/10.1371/journal.pone.0353261
O'Driscoll R, Turicchi J, Beaulieu K, et al. How well do activity monitors estimate energy expenditure? A systematic review and meta-analysis of the validity of current technologies. British Journal of Sports Medicine. 2020. https://doi.org/10.1136/bjsports-2018-099643
Ferguson T, Olds T, Curtis R, et al. Effectiveness of wearable activity trackers to increase physical activity and improve health: a systematic review of systematic reviews and meta-analyses. The Lancet Digital Health. 2022. https://doi.org/10.1016/S2589-7500(22)00111-X
Wyatt KD, Poole LR, Mullan AF, et al. Clinical evaluation and diagnostic yield following evaluation of abnormal pulse detected using Apple Watch. Journal of the American Medical Informatics Association. 2020. https://doi.org/10.1093/jamia/ocaa137
Rosman L, Lampert R, Zhuo S, et al. Wearable Devices, Health Care Use, and Psychological Well-Being in Patients With Atrial Fibrillation. Journal of the American Heart Association. 2024. https://doi.org/10.1161/JAHA.123.033750
Khatsenko K, Khin Y, Maibach H. Allergic Contact Dermatitis to Components of Wearable Adhesive Health Devices. Dermatitis. 2020. https://doi.org/10.1097/DER.0000000000000575

Everything in the library