This paper is in the following e-collection/theme issue:

Fitness Trackers and Smart Pedometers/Accelerometers (571) Pulmonology, COPD (145) Validity, Reliability, and Bias (94) Wearable Devices and Sensors (903) mHealth for Data Collection and Research (981) Innovations and Technology for Physical Activity Education (756)

Published on in Vol 12 (2024)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/56027, first published .
Validity of a Consumer-Based Wearable to Measure Clinical Parameters in Patients With Chronic Obstructive Pulmonary Disease and Healthy Controls: Observational Study

Validity of a Consumer-Based Wearable to Measure Clinical Parameters in Patients With Chronic Obstructive Pulmonary Disease and Healthy Controls: Observational Study

Validity of a Consumer-Based Wearable to Measure Clinical Parameters in Patients With Chronic Obstructive Pulmonary Disease and Healthy Controls: Observational Study

Authors of this article:

Fien Hermans1, 2 Author Orcid Image ;   Eva Arents1 Author Orcid Image ;   Astrid Blondeel2 Author Orcid Image ;   Wim Janssens3, 4 Author Orcid Image ;   Nina Cardinaels5 Author Orcid Image ;   Patrick Calders1 Author Orcid Image ;   Thierry Troosters2 Author Orcid Image ;   Eric Derom5 Author Orcid Image ;   Heleen Demeyer1, 2 Author Orcid Image
Article Authors Cited by Tweetations (8) Metrics

    1Department of Rehabilitation Sciences, Ghent University, , Corneel Heymanslaan 10, Entrance 46, Ghent, , Belgium

    2Department of Rehabilitation Sciences, KU Leuven, , Leuven, , Belgium

    3Department of Chronic Diseases, Metabolism and Aging (CHROMETA) - BREATHE, KU Leuven, , Leuven, , Belgium

    4Clinical Department of Respiratory Diseases, University Hospitals Leuven, , Leuven, , Belgium

    5Clinical Department of Respiratory Medicine, Ghent University Hospital, , Ghent, , Belgium

    Corresponding Author:

    Heleen Demeyer, PhD


    Background: Consumer-based wearables are becoming more popular and provide opportunities to track individual’s clinical parameters remotely. However, literature about their criterion and known-groups validity is scarce.

    Objective: This study aimed to assess the validity of the Fitbit Charge 4, a wrist-worn consumer-based wearable, to measure clinical parameters (ie, daily step count, resting heart rate [RHR], heart rate variability [HRV], respiratory rate [RR], and oxygen saturation) in patients with chronic obstructive pulmonary disease (COPD) and healthy controls in free-living conditions in Belgium by comparing it with medical-grade devices.

    Methods: Participants wore the Fitbit Charge 4 along with three medical-grade devices: (1) Dynaport MoveMonitor for 7 days, retrieving daily step count; (2) Polar H10 for 5 days, retrieving RHR, HRV, and RR; and (3) Nonin WristOX2 3150 for 4 nights, retrieving oxygen saturation. Criterion validity was assessed by investigating the agreement between day-by-day measures of the Fitbit Charge 4 and the corresponding reference devices. Known-groups validity was assessed by comparing patients with COPD and healthy controls.

    Results: Data of 30 patients with COPD and 25 age- and gender-matched healthy controls resulted in good agreement between the Fitbit Charge 4 and the corresponding reference device for measuring daily step count (intraclass correlation coefficient [ICC2,1]=0.79 and ICC2,1=0.85, respectively), RHR (ICC2,1=0.80 and ICC2,1=0.79, respectively), and RR (ICC2,1=0.84 and ICC2,1=0.77, respectively). The agreement for HRV was moderate (healthy controls: ICC2,1=0.69) to strong (COPD: ICC2,1=0.87). The agreement in measuring oxygen saturation in patients with COPD was poor (ICC2,1=0.32). The Fitbit device overestimated the daily step count and underestimated HRV in both groups. While RHR and RR were overestimated in healthy controls, no difference was observed in patients with COPD. Oxygen saturation was overestimated in patients with COPD. The Fitbit Charge 4 detected significant differences in daily step count, RHR, and RR between patients with COPD and healthy controls, similar to those identified by the reference devices, supporting known-groups validity.

    Conclusions: Although the Fitbit Charge 4 shows mainly moderate to good agreement, measures of clinical parameters deviated from the reference devices, indicating that monitoring patients remotely and interpreting parameters requires caution. Differences in clinical parameters between patients with COPD and healthy controls that were measured by the reference devices were all detected by the Fitbit Charge 4.

    JMIR Mhealth Uhealth 2024;12:e56027

    doi:10.2196/56027

    Keywords

    chronic obstructive pulmonary disease; COPD; wearable; Fitbit; clinical parameters; physical activity; validity; observational study; wrist-worn wearable; heart rate; heart rate variability; respiratory rate; oxygen saturation; devices; monitoring 


    In recent years, consumer-based wearables have become increasingly popular due to their relatively low cost, ease of use, and ability to provide real-time feedback on several clinical parameters such as heart rate and step count. In Europe, 26% of 45‐ to 74-year-olds used a wearable device in 2022 [1]. Alongside their rising popularity, wearable technologies also advanced significantly. Contemporary generation wrist-worn wearables use photoplethysmography technology, which measures the volumetric variations of blood circulation via an infrared light [2]. This technology enables to measure multiple parameters of autonomic function (ie, resting heart rate [RHR], heart rate variability [HRV], and respiratory rate [RR]) and oxygen saturation (SpO2) [3].

    Chronic obstructive pulmonary disease (COPD) is the third leading cause of death worldwide [4]. Although COPD is a disease characterized by respiratory symptoms and exercise intolerance related to abnormalities of the airways, the alveoli, or both, COPD seems also to alter the autonomic nervous system [5]. The autonomic nervous system adjusts heart rate, blood pressure, and RR in response to internal and external stimuli [6]. Patients with COPD show an elevated RHR, a reduced HRV and an increased RR compared to healthy controls [5,7]. Besides the autonomic function, SpO2 is a clinical parameter that differs between patients with COPD and healthy controls, with patients with COPD having a lower SpO2 due to alterations in gas exchange and experiencing a sudden drop in SpO2 during exercise and when experiencing an acute exacerbation [4,8]. Both parameters of autonomic function and SpO2 are prognostic markers of mortality and thus important to monitor [5,8,9].

    Wearables potentially enable continuous monitoring of clinical parameters remotely and unobtrusively over a long period of time [10]. Whereas manufacturers specify that consumer-based wearables are not intended for medical purposes as they do not qualify as medical devices, continuously monitoring these parameters would provide information on the management of the patients’ health at home, and investigate the effectiveness of interventions based on measures outside clinical visits in an easy way [11]. Moreover, given that these clinical parameters are linked to the worsening of health in various chronic diseases, it is tempting to actively monitor these parameters remotely, as this may lead to earlier detection of patients’ deterioration or provide an explanation for reduced engagement in physical activity [12-16].

    Previous literature showed that Fitbit wearables are valid devices for monitoring daily step counts in healthy individuals and can be used to monitor patterns of physical activity in patients with COPD [17-19]. However, literature in the healthy population on the validity of Fitbit measuring other clinical parameters is scarce, and no data are available for patients with COPD [18,20-22]. These studies used various devices, including medical-grade devices (such as the ActiGraph GT3X+ and Dynaport MoveMonitor for daily step count, and the Polar H7 for heart rate), as well as gold-standard measurements (such as ECG Holter monitoring for heart rate), to assess the validity of Fitbit wearables.

    The aim of this observational study was to investigate the criterion validity and the known-groups validity of a consumer-based wearable for monitoring physical activity (ie, daily step count) and parameters of autonomic function (ie, RHR, HRV, and RR) in patients with COPD and a reference population consisting of healthy age-matched controls. Furthermore, the criterion validity and known-groups validity of this consumer-based wearable for monitoring SpO2 was examined in patients with COPD.


    Population and Design

    This observational study was nested in a randomized controlled trial investigating the long-term effects of a telecoaching intervention in patients with COPD (NCT04139200). All patients with COPD included via Ghent University Hospital (Ghent, Belgium) and examined between April 2022 and June 2023 were enrolled in this substudy, ensuring representation of the entire patient cohort at this site. Patients aged 40 years and older with a smoking history of at least 10 pack years and with a clinical diagnosis of COPD (confirmed by spirometry [Tiffeneau-index<70%]) but no history of exacerbations in the past month were eligible to participate. Patients were excluded if they had orthopedic or other problems preventing them from improving physical activity, had undergone lung transplantation, were involved in a multidisciplinary rehabilitation program, or were unable to learn to work with electronic devices.

    In addition, healthy controls were recruited between November 2022 and August 2023. These participants were 50‐80 years old, had never smoked or had stopped smoking more than 20 years before inclusion, had a Tiffeneau-index ≥70%, had no chronic health problems or orthopedic problems preventing them from being physically active, and did not participate in a rehabilitation program.

    Ethical Considerations

    This study consisted of a single clinical visit and a follow-up period of 7 days. Both studies were approved by the ethical committee of Ghent University Hospital (BC-10267 [COPD] and ONZ-2022‐0387 [healthy controls]). All participants signed the informed consent prior to data collection and were assigned a study ID number. No compensation was provided for participation in this study.

    Clinical Assessments

    At the clinical visit, the following assessments were performed in both groups: (1) sociodemographic and clinical data (age, sex, height, weight, smoking history, and medication intake); (2) postbronchodilator spirometry according to European Respiratory Society (ERS) – American Thoracic Society (ATS) guidelines [23]; (3) functional exercise capacity based on the best of two 6-minute walk tests following ERS-ATS guidelines [24]; and (4) Composite Autonomic Symptom Score (COMPASS) 31 questionnaire questioning autonomic nervous system symptoms [25]. In patients with COPD, the modified Medical Research Council scale (mMRC) questioning dyspnea and COPD Assessment Test (CAT) questioning health status were collected additionally [26,27].

    During the 7-day wearing period following the clinical visit, participants were asked to wear a wrist-worn wearable, the Fitbit Charge 4, along with 3 approved medical-grade devices. During this wearing period, participants completed a daily diary including the time of waking up and going to bed, intake of medication, and consumption of beverages influencing the heart rate (eg, coffee, tea, alcohol, and energy drinks).

    Wearable

    The Fitbit Charge 4 (Fitbit Inc) is a triaxial consumer-based wearable worn on the nondominant wrist of the participant. This device records daily step count (accelerometry), as well as RHR, HRV, RR, and SpO2 (photoplethysmography). These variables were extracted as day-by-day outcomes from the Fitbit platform, as calculated by the proprietary algorithms. Fitbit defines RHR as the heart rate while in a relaxed state during both sleep and being awake. HRV and RR are defined as the variation of duration between heartbeats, expressed as the root mean square of successive differences (RMSSD) in milliseconds (ms), and the number of breaths per minute during the night, respectively, when more than 3 hours of continuous sleep are recorded. Transcutaneous SpO2 was defined as the average hemoglobin SpO2 during the night (>3 hours of continuous sleep). Participants were instructed to wear the device for 7 consecutive days (24 hours per day). The device had a battery life of up to 7 days, but participants were advised to charge the battery when the battery level dropped below 10% and when they were not performing any activity.

    Reference Devices

    Physical Activity—Dynaport MoveMonitor

    The Dynaport MoveMonitor (DAM; McRoberts) is a triaxial accelerometer validated to objectively measure physical activity in patients with COPD and healthy controls [28,29]. Participants were instructed to wear the monitor at the lower back during waking hours for 7 consecutive days, except for bathing and water activities, according to current recommendations [30]. Days with a wearing time lower than 8 hours were excluded for further analyses [30]. Wearing time and daily step count were extracted from the monitor for further analysis.

    Cardiac Autonomic Function—Polar H10

    The Polar H10 sensor chest strap (Polar Electro Oy) is a validated device to capture heart rate and HRV in healthy subjects [31]. This sensor was moistened before being applied below the chest muscles of the participants, as described by the manufacturer. To collect heart rate data for 100 hours, participants wore a corresponding Polar Ignite 2 watch (at the preferred wrist), which was used as a Polar H10 data logger. They were instructed to wear these devices for 5 consecutive days (24 hours per day), except for bathing and water activities, based on recommendations and the maximum recording capacity of 100 hours. Data were recorded using a 1-second time interval. Heart rate and beat-to-beat RR intervals (time between 2 successive R-waves of the QRS signal on the electrocardiogram) were extracted from the device using the Polar Flow web service.

    To obtain the RHR, the average heart rate during the night (actual sleep time, ie, starting 1 hour later than the reported time of going to bed and ending 1 hour earlier than the reported time of waking up) was calculated to best match Fitbit’s definition. Nights with 3 hours of sleep or less were excluded for further analyses.

    The beat-to-beat RR intervals (time between 2 successive R-waves of the QRS signal on the electrocardiogram) were transferred to Kubios HRV Software (version 4.0.1, Kuopio) to analyze HRV (expressed as RMSSD in ms) and RR [32]. The actual sleep time as reported in the diary was entered into the software. Nights with 3 hours of sleep or less were excluded for further analyses to best fit with Fitbit’s definition.

    SpO2 (COPD Only)—Nonin WristOX2 3150

    The wearable finger pulse oximeter Nonin WristOX2 3150 (Nonin Medical Inc) was used as a reference device for measuring SpO2, as it has become a commonly used device for home monitoring of patients with COPD and has a high accuracy (±2%) according to the manufacturer [33]. Patients with COPD were instructed to wear the device on the index finger of their dominant hand for 4 consecutive nights (limited battery life of 48 hours using continuous measurements, allowing up to 12 hours of monitoring per night). Data were recorded using a 4-second time interval. Data were stored on the internal memory of the device and downloaded after the 7-day follow-up period using the nVISION software (version 6.5.1.2, Nonin Medical Inc). The average SpO2 during the night measured by Nonin was calculated after removing impossible values. Nights with 3 hours of sleep (as judged by the Fitbit device) or less were excluded for further analyses to best fit with Fitbit’s definition.

    Statistical Analyses

    The sample size of this substudy was chosen to align with prior research within this field [18]. All statistical analyses were performed using the SAS statistical package (version 9.4, SAS Institute). Data are presented as mean (SD) or median (IQR), as appropriate after testing for normality using the Shapiro-Wilk test. Statistical significance was set at P<.05 for all analyses.

    Criterion validity was investigated by comparing day-by-day data obtained by the Fitbit Charge 4 with the reference devices by use of a 2-tailed paired t test, Bland-Altman plots, and intraclass correlation coefficients (ICC2,1). The cutoffs that were used to interpret the findings were ICC<0.50 as “poor,” ICC=0.50‐0.75 as “moderate,” ICC=0.75‐0.90 as “good,” and ICC>0.90 as “excellent” [34]. These analyses were performed for patients with COPD and healthy controls separately. To evaluate whether the Fitbit Charge 4 is able to pick up day-by-day fluctuations, the delta (day minus day-1) of each measured clinical parameter was calculated on consecutive days. The agreement between the Fitbit Charge 4 and the corresponding reference device with regard to day-by-day fluctuations was determined via Pearson correlation. The correlation was interpreted using the cutoffs r<0.30 classed as “no correlation,” r=0.30‐0.50 as “weak correlation,” r=0.50‐0.70 as “moderate correlation,” r=0.70‐0.90 as “strong correlation,” and r>0.90 as “very strong correlation” [35]. Next, known-groups validity was assessed by investigating whether the differences between patients with COPD and healthy controls were picked up by the reference device and the Fitbit Charge 4 in the same way. For this, the outcomes were compared between patients with COPD and healthy controls for both devices using an unpaired t test. The differences between both groups were examined by an interaction effect based on a mixed model analysis. As a sensitivity analysis, all analyses were performed with the exclusion of patients taking beta-blockers.


    Patient Characteristics

    In total, 32 patients with COPD and 26 age- and gender-matched healthy controls were included in this study. Overall, valid data of 30 patients with COPD (1 patient dropped out and 1 patient was not willing to wear the additional devices) and 25 healthy controls (1 participant was excluded because spirometry displayed an obstructive syndrome) were obtained, but across the clinical parameters, there is a variation in participants included in the analyses. This variation can be attributed to various reasons, such as technical problems (the Fitbit not capturing clinical parameters at night, sudden interruptions in the measurement by the reference device, battery issues), as well as participants forgetting to wear the (reference) device. Baseline characteristics are shown in Table 1.

    Table 1. Characteristics of all participants included in the analyses.
    Patients with COPDa (n=30)Healthy controls (n=25)P value
    Age (years), mean (SD)70 (7)68 (7).28
    Sex (male), n (%)23 (77)20 (80).77
    BMI (kg/m2), mean (SD)27 (6)28 (4).86
    Forced expiratory volume in the first second (% predicted), mean (SD)54 (15)109 (15)<.001
    Severity of COPD, n (%) 
    I-II2 (7)-15 (50)N/AbN/A
    III-IV12 (40)-1 (3)N/AN/A
    Use of β-blockers, n (%)8 (27)4 (16).34
    Current smokers, n (%)1 (3)0 (0).36
    6MWDc (m), mean (SD)482 (95)618 (81)<.001
    6MWD (% predicted), mean (SD)78 (14)97 (10)<.001
    COMPASS 31d (score), median (IQR)11 (4‐21)5 (2-12).005
    mMRCe (score), median (IQR)2 (1-2)N/AN/A
    CATf (score), mean (SD)15 (6)N/AN/A

    aCOPD: chronic obstructive pulmonary disease.

    bN/A: not applicable.

    c6MWD: 6-minute walking distance.

    dCOMPASS 31: Composite Autonomic Symptom Score 31 questionnaire (0‐100); a higher score indicated more symptoms of autonomic dysfunction.

    emMRC: modified Medical Research Council dyspnea scale (0‐4); a higher score indicated more dyspnea.

    fCAT: COPD Assessment Test (0‐40); a higher score indicated a worse health status.

    Criterion Validity

    Physical Activity

    Daily step count was analyzed based on 199 (min-max 4 to 7 days per patient, n=30) and 157 (4 to 7 days per participant, n=24) overlapping data points for patients with COPD and healthy controls, respectively. As expected, patients with COPD were less active compared to healthy controls (Table 2).

    Table 2. Average step count, resting heart rate, heart rate variability, and respiratory rate in patients with COPDa and healthy controls.
    Patients with COPD (n=30), mean (SD)Healthy controls (n=25), mean (SD)Difference, mean (SD)P valueb
    Daily steps (steps/day)   
    Dynaport MoveMonitor5625 (3618)7933 (4263)−2308 (3915)<.001
    Fitbit Charge 47423 (4325)9055 (4956)−1632 (4613).001
    Resting heart rate (beats/min)   
    Polar H1070 (8)60 (10)10 (9)<.001
    Fitbit Charge 470 (8)64 (10)6 (9)<.001
    Heart rate variability (ms)   
    Polar H1027 (19)33 (21)−6 (20).16
    Fitbit Charge 424 (14)26 (13)−2 (14).53
    Respiratory rate (breaths/min)   
    Polar H1016 (2)14 (2)2 (2)<.001
    Fitbit Charge 416 (3)15 (2)1 (2).01

    aCOPD: chronic obstructive pulmonary disease.

    bP values based on unpaired t test.

    In both groups, the mean step count measured by the Fitbit Charge 4 was significantly higher compared to DAM (mean, SD; COPD: ∆+1798, SD 2070 steps/day; P<.001 and healthy controls: ∆+1122, SD 2297 steps/day; P<.001). However, in both groups, a good agreement between the devices was found (COPD: ICC2,1=0.79; 95% CI 0.36‐0.90 and healthy controls: ICC2,1=0.85; 95% CI 0.74‐0.91). These findings are supported by the Bland-Altman plots depicted in Figure 1A and 1B. Concurrent validity with associated scatterplots can be found in Multimedia Appendix 1.

    ‎
    Figure 1. Bland-Altman plots with mean and 95% CI for Fitbit Charge 4 compared to the reference devices. (A, B) Daily steps (steps/day) measured by Fitbit Charge 4 and DAM (COPD: n=30 and healthy controls: n=24). (C, D) Resting heart rate (beats/min) measured by Fitbit Charge 4 and Polar H10 (COPD: n=25 and healthy controls: n=20). (E, F) Heart rate variability (RMSSD in ms) measured by Fitbit Charge 4 and Polar H10 (COPD: n=22 and healthy controls: n=16). (G, H) Respiratory rate (breaths/min) measured by Fitbit Charge 4 and Polar H10 (COPD: n=21 and healthy controls: n=17). (I) Oxygen saturation (%) measured by Fitbit Charge 4 and Nonin WristOX2 3150 (COPD: n=19). Large open dots represent the mean individual outcome per patient. Small dots represent daily data. Mean and 95% CI are calculated based on average data. COPD: chronic obstructive pulmonary disease; DAM: Dynaport MoveMonitor; RMSSD: root mean square of successive differences.

    Figure 2A shows the scatterplot of the day-by-day fluctuations in daily step count measured with the Fitbit Charge 4 and the DAM. A strong to very strong association was observed between both devices in patients with COPD (r=0.71) and healthy controls (r=0.91), respectively.

    ‎
    Figure 2. Scatterplots of the day-by-day fluctuations in daily step count, parameters of autonomic function, and SpO2 in patients with COPD and healthy controls. (A) Day-by-day fluctuations in daily step count (COPD: 169 data points and healthy controls: 133 data points). (B) Day-by-day fluctuations in resting heart rate (COPD: 42 data points and healthy controls: 39 data points). (C) Day-by-day fluctuations in heart rate variability (COPD: 27 data points and healthy controls: 28 data points). (D) Day-by-day fluctuations in respiratory rate (COPD: 24 data points and healthy controls: 30 data points). (E) Day-by-day fluctuations in oxygen saturation (COPD: 42 data points). Patients with COPD are depicted in the closed dots and healthy controls are depicted in the open dots. An identity line is displayed on the scatterplots. COPD: chronic obstructive pulmonary disease; DAM: Dynaport MoveMonitor; HRV: heart rate variability; RHR: resting heart rate; RR: respiratory rate; SpO2: oxygen saturation.
    Resting Heart Rate

    RHR was analyzed based on 66 (min-max 1 to 4 days per patient, n=25) and 59 (1 to 4 days per healthy control, n=20) overlapping data points. RHR was higher in patients with COPD compared to healthy controls (Table 2).

    In the COPD group, RHR measured by the Fitbit showed no difference compared to Polar H10 (∆−0.3, SD 5 beats/min; P=.67), whereas Fitbit significantly overestimated RHR in the healthy control group (∆+4, SD 6 beats/min; P<.001). In both groups, Fitbit Charge 4 and Polar H10 showed a good agreement for assessing RHR (COPD: ICC2,1=0.80; 95% CI 0.70‐0.87 and healthy controls: ICC2,1=0.79; 95% CI 0.50‐0.90). The findings are depicted in Bland-Altman plots (Figure 1C and 1D). A strong association between the devices was observed in both groups (Multimedia Appendix 1).

    A weak negative association was observed in patients with COPD for picking up day-by-day fluctuations in RHR (r=−0.35; Figure 2B).

    Heart Rate Variability

    HRV analyses are based on 49 (min-max 1 to 3 days per patient, n=22) and 44 (1 to 4 days per participant, n=16) overlapping data points for patients with COPD and healthy controls, respectively. No difference in HRV was observed between patients with COPD and healthy controls (Table 2).

    The Fitbit Charge 4 significantly underestimated HRV in both groups (respectively in COPD and healthy controls: ∆−3, SD 8 ms; P=.03 and ∆−7, SD 13 ms; P=.001). In patients with COPD, a good agreement was observed between both devices (ICC2,1=0.87; 95% CI 0.77‐0.93). Whereas in the healthy control group, a moderate agreement was found for measuring HRV (ICC2,1=0.69; 95% CI 0.43‐0.83). Bland-Altman plots (Figure 1E and 1F) represent these results. Scatterplots showing the concurrent validity can be found in Multimedia Appendix 1.

    In patients with COPD, a strong association was found for picking up day-by-day fluctuations in HRV (r=0.83), whereas healthy controls exhibited a moderate association (r=0.64; Figure 2C).

    Respiratory Rate

    RR was analyzed based on 45 (min-max 1 to 5 days per patient, n=21) and 47 (1 to 4 days per healthy control, n=17) overlapping data points. RR was higher in patients with COPD compared to healthy controls (Table 2).

    In patients with COPD, no difference was detected in RR measured by Fitbit Charge 4 compared to Polar H10 (∆+0.3, SD 1 breaths/min; P=.25). However, in the healthy control group, the Fitbit Charge 4 significantly overestimated the RR (∆+1, SD 1 breaths/min; P<.001). A good agreement was observed in patients with COPD, as well as healthy controls (COPD: ICC2,1=0.84; 95% CI 0.72‐0.91 and healthy controls: ICC2,1=0.77; 95% CI 0.44‐0.89). These findings are supported by the Bland-Altman plots depicted in Figure 1G and 1H. In both groups, a strong association was found (Multimedia Appendix 1).

    A weak association was observed in both groups for picking up day-by-day fluctuations in RR (COPD: r=0.47 and healthy controls: r=0.50; Figure 2D).

    Oxygen Saturation

    SpO2 was analyzed based on 61 (min-max 1 to 5 days per patient, n=19) overlapping data points in patients with COPD. Fitbit Charge 4 significantly overestimated SpO2 (∆+2, SD 2%; P<.001). A poor agreement between both devices was found (ICC2,1=0.32; 95% CI −0.10‐0.65). The Bland-Altman analysis is shown in Figure 1I.

    Figure 2E displays the scatterplot of the day-by-day fluctuations measured with the Fitbit Charge 4 and the Nonin WristOX2 3150, showing a weak association for SpO2 in patients with COPD (r=0.40).

    Known-Groups Validity

    The significant differences between patients with COPD and healthy controls in daily steps, RHR, and RR identified by the reference devices were picked up in a similar way by the Fitbit Charge 4 (see Table 2). The difference between both groups is significantly smaller when assessing RHR or RR using the Fitbit Charge 4 compared to the reference device (P<.05).

    Sensitivity Analysis

    Excluding participants on stable doses of beta-blockers had minimal impact on the results. The agreement between the Fitbit Charge 4 and its corresponding reference devices remained unchanged. The exclusion of participants taking beta-blockers had no effect on the known-groups validity (Multimedia Appendix 2).


    Principal Findings

    This study, which aimed to investigate the validity of a consumer-based wearable, the Fitbit Charge 4, in patients with COPD and healthy controls found mixed results for criterion validity. First, the Fitbit Charge 4 significantly overestimated daily step count and significantly underestimated HRV in patients with COPD and healthy controls. In patients with COPD, RHR and RR are not different between the Fitbit Charge 4 and the Polar H10, but both parameters were overestimated in the healthy control group. The Fitbit Charge 4 significantly overestimated SpO2 in patients with COPD. Second, the agreement between the Fitbit Charge 4 and the corresponding reference devices is moderate to good for monitoring most clinical parameters (ie, daily steps, RHR, HRV, and RR) but poor for tracking SpO2. The Fitbit is able to pick up day-by-day fluctuations in daily step count and HRV but lacks accuracy to pick up the small day-by-day fluctuations in RHR, RR, and SpO2. The known-groups validity of the Fitbit Charge 4 is good. All expected differences between patients with COPD and age- and gender-matched healthy controls are picked up by the wearable.

    Our results are consistent with previous research in older adults in free-living conditions showing that the daily step count measured by the Fitbit Charge, a wrist-worn wearable, is highly correlated with the daily step count measured by a validated accelerometer, but the Fitbit Charge significantly overestimates the daily step count [21,36,37]. Previous studies conducted in other patient cohorts within our department showed that the wrist-worn Fitbit device significantly overestimated the daily step count in healthy individuals and cancer survivors, whereas it did not in people with Parkinson disease [38,39]. Blondeel et al [19] concluded that the Fitbit Alta, also a wrist-worn wearable, did not significantly overestimate step count in patients with COPD, but did in healthy controls. This corresponds well with our findings, although in our sample, the wearable also overestimated daily step count in patients with COPD. This could potentially be explained by the inclusion of patients with a better functional exercise tolerance (6-minute walking distance of 482 m vs 454 m) in this study.

    A few studies examined the validity of Fitbit measuring the heart rate during sleep among healthy adults in different situations (ie, home environment and laboratory-based setting). One study in healthy adolescents showed that Fitbit significantly underestimated the heart rate by 0.9 beats/min on average as compared to ECG, while another study in healthy adults found no difference between Fitbit and the reference device [40,41]. These findings are consistent with our results in the COPD group but are inconsistent with our results showing that Fitbit significantly overestimates RHR by 4 beats/min in the healthy control group. The discrepancy in results can possibly be attributed to the difference in how the definition is applied, with Fitbit defining RHR as the heart rate while in a relaxed state during both sleep and being awake, whereas we calculated RHR from Polar for sleep time values only because we do not have exact information on how RHR is calculated by Fitbit.

    To the best of our knowledge, the validity of Fitbit measuring HRV, RR, and SpO2 has not been investigated so far. However, some research has been performed regarding other wearables also using photoplethysmography technology (eg, Apple watch, Garmin, and Polar), showing a large range for measuring HRV (ICC ranging from 0.24 to 0.99) depending on the wearable used in a laboratory setting in healthy adults [42]. These results are consistent with our findings in healthy controls (ICC2,1=0.69). Interestingly, we found a higher agreement in patients with COPD (ICC2,1=0.87), but no previous research has been conducted on the validity of wearables measuring HRV among this population.

    Data on the validity of a wrist-worn wearable estimating RR based on photoplethysmography have not been published so far. Existing studies have only focused on the validity of devices measuring this parameter using different technologies (ie, wearable biosensors and wearable pressure sensors) [43,44]. This difference in technology makes it challenging to compare these results.

    Several studies indicate that the Apple Watch Series 6 shows a good agreement for measuring SpO2 compared to the gold-standard (ie, arterial blood gas analyses) or reference devices (ie, oximeters) in patients with lung diseases (ICCs ranging from 0.90 to 0.94), as well as healthy individuals (mean bias 1.7, SD 2.1%) [45-47]. These results are in contrast with our findings achieving an ICC of 0.32. The results of Garmin concerning the measurements of SpO2 in healthy individuals vary widely, with ICCs ranging from 0.28 to 0.55, which is more in line with our results [48,49]. In our study, the SpO2 was measured during sleep, where the contact between the watch and the skin was maybe occasionally suboptimal due to unintentionally lying on the wearable, leading to divergent measurements through photoplethysmography. This is in contrast to the other studies where the measurement of SpO2 was conducted under controlled settings over a short period of time (maximum 10 minutes) while being awake.

    Regarding the known-groups validity, the differences in daily step count (lower in COPD), RHR (higher in COPD), and RR (higher in COPD) were in line with previous research [7,9,19]. We could not confirm a difference in HRV, which was shown to be lower in COPD in previous research (COPD and healthy controls: 27, SD 19 ms and 33, SD 21 ms; P=.16 [this study] vs 11, SD 3 ms and 19, SD 7 ms; P=.002, respectively) [50]. This study confirms that previously laboratory-based findings are now also observed in free-living situations.

    It is remarkable that our findings indicate that Fitbit estimates certain clinical parameters, such as RHR, HRV, and RR, better in patients with COPD as compared to healthy controls. In this regard, we speculate that faster aging of the skin and skin thinning, a typical feature in patients with COPD on inhaled and systemic steroids, improve reflection of lights for measuring clinical parameters through photoplethysmography compared to slightly thicker skins in healthy controls as the distance of the infrared light to measure volumetric changes in the blood is reduced [51-53].

    Clinical Implications

    While our results suggest that Fitbit can measure certain clinical parameters better in patients with COPD than in healthy controls, it is not the best choice if one aims to measure important parameters of autonomic function (ie, RHR and RR) as well as SpO2 in patients with COPD, as the wearable fails to pick up day-by-day fluctuations. Nevertheless, Fitbit can be used for commercial purposes and well-being monitoring, including data on daily step counts and HRV. Advanced and accurate (medical) devices are more appropriate for remotely monitoring clinical parameters. However, these results are only based on a 5‐ to 7-day assessment in a stable situation. Longer time series of data also including larger fluctuations in the clinical parameters (eg, when patients experience a deterioration in their health) are needed to confirm these findings.

    Strengths and Limitations

    This study offers valuable insights into the validity of a popular wearable device to estimate various clinical parameters during a free-living situation. Our approach involved studying both patients with COPD, as well as age- and gender-matched healthy controls, aiming to establish evidence in both populations. Even though 4 of the 25 (16%) control participants were using beta-blockers, they can be considered as a representative sample of the general population [54]. Nonetheless, this study has some limitations. First, it should be mentioned that Fitbit operates as a black box, using proprietary algorithms of which its details are not disclosed. This makes it challenging to establish an agreement with reference devices. Second, technical issues arose, preventing the collection of certain data. In some cases, Fitbit failed to measure parameters of autonomic function or SpO2, rendering a comparison with reference devices impossible. Third, validated medical-grade devices that show good agreement with gold-standard methods were used in this study to assess the validity of the Fitbit Charge 4 in free-living conditions. Unfortunately, gold-standard devices (ie, video recording or manual step counting [physical activity] and ECG Holter monitoring [HRV]) are restricted to laboratory-based measurements. However, it is important to acknowledge that the reference devices may still involve measurement inaccuracies. Given that the measurements were conducted in daily life, the results have a high level of generalizability. Fourth, a selection bias within our healthy controls is difficult to avoid, as probably the more motivated individuals are more prone to participate in these kinds of studies. However, the clinical parameters in our control group align with those reported in the general population [55,56].

    Conclusions

    Both in patients with COPD and healthy controls, measures of clinical parameters collected by the commercially Fitbit Charge 4 showed moderate to good agreement with the reference devices. However, these measures deviated significantly. In patients with COPD, the Fitbit Charge 4 is accurate in measuring RHR and RR. The wearable lacks accuracy to pick up day-by-day fluctuations in RHR, RR, and SpO2; hence, the Fitbit Charge 4 should be used with caution when information on clinical parameters is collected over a short period of time.

    Acknowledgments

    The authors would like to thank the colleagues from the pulmonary function testing at Ghent University Hospital for their assistance with the clinical assessments. They would also like to thank Dr Nathalie Michels for providing access to the Kubios software and for sharing her expertise on its usage. This research was funded by the Flemish Research Foundation (FWO project G0C0720N). AB is a predoctoral research fellow of the Flemish Research Foundation (FWO-Flanders 1194320N). WJ is a senior clinical research fellow of the Flemish Research Foundation (FWO-Flanders). HD is a postdoctoral research fellow of the Flemish Research Foundation (FWO-Flanders 12H7517N).

    Conflicts of Interest

    None declared.

    Multimedia Appendix 1

    Concurrent validity between the Fitbit Charge 4 and the corresponding reference device for measuring daily step count, resting heart rate, heart rate variability, and respiratory rate.

    DOCX File, 358 KB
    Multimedia Appendix 2

    Sensitivity analysis—criterion and known-groups validity of the Fitbit Charge 4 in participants not taking beta-blockers.

    DOCX File, 21 KB
    1. Growing importance of internet-connected devices. Eurostat. URL: https://ec.europa.eu/eurostat/web/products-eurostat-news/w/ddn-20230829-1 [Accessed 2023-08-25]
    2. Tamura T, Maeda Y, Sekine M, Yoshida M. Wearable photoplethysmographic sensors—past and present. Electronics (Basel). 2014;3(2):282-302. [CrossRef]
    3. Castaneda D, Esparza A, Ghamari M, Soltanpur C, Nazeran H. A review on wearable photoplethysmography sensors and their potential future applications in health care. Int J Biosens Bioelectron. 2018;4(4):195-202. [CrossRef] [Medline]
    4. Global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease: 2024 report. Global Initiative for Chronic Obstructive Lung Disease. 2024. URL: https://goldcopd.org/2024-gold-report/ [Accessed 2023-08-02]
    5. van Gestel AJR, Steier J. Autonomic dysfunction in patients with chronic obstructive pulmonary disease (COPD). J Thorac Dis. Dec 2010;2(4):215-222. [CrossRef] [Medline]
    6. Wehrwein EA, Orer HS, Barman SM. Overview of the anatomy, physiology, and pharmacology of the autonomic nervous system. Compr Physiol. Jun 13, 2016;6(3):1239-1278. [CrossRef] [Medline]
    7. Yentes JM, Fallahtafti F, Denton W, Rennard SI. COPD patients have a restricted breathing pattern that persists with increased metabolic demands. COPD. Jun 2020;17(3):245-252. [CrossRef] [Medline]
    8. Kent BD, Mitchell PD, McNicholas WT. Hypoxemia in patients with COPD: cause, effects, and disease progression. Int J Chron Obstruct Pulmon Dis. 2011;6:199-208. [CrossRef] [Medline]
    9. Jensen MT, Marott JL, Lange P, et al. Resting heart rate is a predictor of mortality in COPD. Eur Respir J. Aug 2013;42(2):341-349. [CrossRef] [Medline]
    10. Buekers J, Arbillaga-Etxarri A, Gimeno-Santos E, et al. Heart rate and oxygen uptake kinetics obtained from continuous measurements with wearable devices during outdoor walks of patients with COPD. Digit Health. 2023;9:20552076231162989. [CrossRef] [Medline]
    11. Fitbit important safety and product and services information. Fitbit device documentation. URL: https://www.fitbit.com/global/us/legal/safety-instructions [Accessed 2023-11-08]
    12. Nolan J, Flapan AD, Capewell S, MacDonald TM, Neilson JM, Ewing DJ. Decreased cardiac parasympathetic activity in chronic heart failure and its relation to left ventricular function. Br Heart J. Jun 1992;67(6):482-485. [CrossRef] [Medline]
    13. De Pablo-Fernandez E, Tur C, Revesz T, Lees AJ, Holton JL, Warner TT. Association of autonomic dysfunction with disease progression and survival in Parkinson disease. JAMA Neurol. Aug 1, 2017;74(8):970-976. [CrossRef] [Medline]
    14. Lanza GA, Guido V, Galeazzi MM, et al. Prognostic role of heart rate variability in patients with a recent acute myocardial infarction. Am J Cardiol. Dec 1998;82(11):1323-1328. [CrossRef]
    15. Raju S, Woo H, Fawzy A, et al. Decreased cardiac autonomic function is associated with higher exacerbation risk and symptom burden in chronic obstructive pulmonary disease. Chronic Obstr Pulm Dis. Jul 26, 2023;10(3):328-334. [CrossRef] [Medline]
    16. Hurst JR, Donaldson GC, Quint JK, Goldring JJ, Patel AR, Wedzicha JA. Domiciliary pulse-oximetry at exacerbation of chronic obstructive pulmonary disease: prospective pilot study. BMC Pulm Med. Dec 2010;10(1):52. [CrossRef]
    17. Paul SS, Tiedemann A, Hassett LM, et al. Validity of the Fitbit activity tracker for measuring steps in community-dwelling older adults. BMJ Open Sport Exerc Med. 2015;1(1):e000013. [CrossRef] [Medline]
    18. 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 Uhealth. Sep 8, 2020;8(9):e18694. [CrossRef] [Medline]
    19. Blondeel A, Demeyer H, Janssens W, Troosters T. Accuracy of consumer-based activity trackers as measuring tool and coaching device in patients with COPD and healthy controls. PLoS One. 2020;15(8):e0236676. [CrossRef] [Medline]
    20. Nissen M, Slim S, Jäger K, et al. Heart rate measurement accuracy of Fitbit Charge 4 and Samsung Galaxy Watch Active2: device evaluation study. JMIR Form Res. Mar 1, 2022;6(3):e33635. [CrossRef] [Medline]
    21. Bai Y, Hibbing P, Mantis C, Welk GJ. Comparative evaluation of heart rate-based monitors: Apple Watch vs Fitbit Charge HR. J Sports Sci. Aug 2018;36(15):1734-1741. [CrossRef] [Medline]
    22. Benedetto S, Caldato C, Bazzan E, Greenwood DC, Pensabene V, Actis P. Assessment of the Fitbit Charge 2 for monitoring heart rate. PLoS One. 2018;13(2):e0192691. [CrossRef] [Medline]
    23. Graham BL, Steenbruggen I, Miller MR, et al. Standardization of spirometry 2019 update. An official American Thoracic Society and European Respiratory Society technical statement. Am J Respir Crit Care Med. Oct 15, 2019;200(8):e70-e88. [CrossRef] [Medline]
    24. Holland AE, Spruit MA, Troosters T, et al. An official European Respiratory Society/American Thoracic Society technical standard: field walking tests in chronic respiratory disease. Eur Respir J. Dec 2014;44(6):1428-1446. [CrossRef] [Medline]
    25. Sletten DM, Suarez GA, Low PA, Mandrekar J, Singer W. COMPASS 31: a refined and abbreviated Composite Autonomic Symptom Score. Mayo Clin Proc. Dec 2012;87(12):1196-1201. [CrossRef] [Medline]
    26. Mahler DA, Wells CK. Evaluation of clinical methods for rating dyspnea. Chest. Mar 1988;93(3):580-586. [CrossRef] [Medline]
    27. Jones PW, Harding G, Berry P, Wiklund I, Chen WH, Kline Leidy N. Development and first validation of the COPD assessment test. Eur Respir J. Sep 2009;34(3):648-654. [CrossRef] [Medline]
    28. Van Remoortel H, Raste Y, Louvaris Z, et al. Validity of six activity monitors in chronic obstructive pulmonary disease: a comparison with indirect calorimetry. PLoS One. 2012;7(6):e39198. [CrossRef] [Medline]
    29. Rabinovich RA, Louvaris Z, Raste Y, et al. Validity of physical activity monitors during daily life in patients with COPD. Eur Respir J. Nov 2013;42(5):1205-1215. [CrossRef] [Medline]
    30. Demeyer H, Mohan D, Burtin C, et al. Objectively measured physical activity in patients with COPD: recommendations from an international task force on physical activity. Chronic Obstr Pulm Dis. Oct 28, 2021;8(4):528-550. [CrossRef] [Medline]
    31. Gilgen-Ammann R, Schweizer T, Wyss T. RR interval signal quality of a heart rate monitor and an ECG Holter at rest and during exercise. Eur J Appl Physiol. Jul 2019;119(7):1525-1532. [CrossRef] [Medline]
    32. Tarvainen MP, Niskanen JP, Lipponen JA, Ranta-Aho PO, Karjalainen PA. Kubios HRV--heart rate variability analysis software. Comput Methods Programs Biomed. 2014;113(1):210-220. [CrossRef] [Medline]
    33. Buekers J, Theunis J, De Boever P, et al. Wearable finger pulse oximetry for continuous oxygen saturation measurements during daily home routines of patients with chronic obstructive pulmonary disease (COPD) over one week: observational study. JMIR Mhealth Uhealth. Jun 6, 2019;7(6):e12866. [CrossRef] [Medline]
    34. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med. Jun 2016;15(2):155-163. [CrossRef] [Medline]
    35. Mukaka MM. Statistics corner: a guide to appropriate use of correlation coefficient in medical research. Malawi Med J. Sep 2012;24(3):69-71. [Medline]
    36. Tedesco S, Sica M, Ancillao A, Timmons S, Barton J, O’Flynn B. Validity evaluation of the Fitbit Charge2 and the Garmin vivosmart HR+ in free-living environments in an older adult cohort. JMIR Mhealth Uhealth. Jun 19, 2019;7(6):e13084. [CrossRef] [Medline]
    37. Farina N, Lowry RG. The validity of consumer-level activity monitors in healthy older adults in free-living conditions. J Aging Phys Act. Jan 1, 2018;26(1):128-135. [CrossRef] [Medline]
    38. De Groef A, Asnong A, Blondeel A, et al. Accuracy of consumer-based activity trackers as measuring tool and coaching device in breast and colorectal cancer survivors. Support Care Cancer. Sep 28, 2023;31(10):596. [CrossRef] [Medline]
    39. Ginis P, Goris M, De Groef A, et al. Validation of commercial activity trackers in everyday life of people with Parkinson’s disease. Sensors (Basel). Apr 21, 2023;23(8):4156. [CrossRef] [Medline]
    40. de Zambotti M, Baker FC, Willoughby AR, et al. Measures of sleep and cardiac functioning during sleep using a multi-sensory commercially-available wristband in adolescents. Physiol Behav. May 1, 2016;158:143-149. [CrossRef] [Medline]
    41. Haghayegh S, Khoshnevis S, Smolensky MH, Diller KR. Accuracy of PurePulse photoplethysmography technology of Fitbit Charge 2 for assessment of heart rate during sleep. Chronobiol Int. Jul 2019;36(7):927-933. [CrossRef] [Medline]
    42. Miller DJ, Sargent C, Roach GD. A validation of six wearable devices for estimating sleep, heart rate and heart rate variability in healthy adults. Sensors (Basel). Aug 22, 2022;22(16):6317. [CrossRef] [Medline]
    43. Breteler MJMM, Huizinga E, van Loon K, et al. Reliability of wireless monitoring using a wearable patch sensor in high-risk surgical patients at a step-down unit in the Netherlands: a clinical validation study. BMJ Open. Feb 27, 2018;8(2):e020162. [CrossRef] [Medline]
    44. Takahashi Y, Okura K, Minakata S, et al. Accuracy of heart rate and respiratory rate measurements using two types of wearable devices. Prog Rehabil Med. 2022;7:20220016. [CrossRef] [Medline]
    45. Pipek LZ, Nascimento RFV, Acencio MMP, Teixeira LR. Comparison of SpO2 and heart rate values on Apple Watch and conventional commercial oximeters devices in patients with lung disease. Sci Rep. Sep 23, 2021;11(1):18901. [CrossRef] [Medline]
    46. Arslan B, Sener K, Guven R, et al. Accuracy of the Apple Watch in measuring oxygen saturation: comparison with pulse oximetry and ABG. Ir J Med Sci. Feb 2024;193(1):477-483. [CrossRef] [Medline]
    47. Hearn EL, Byford J, Wolfe C, et al. Measuring arterial oxygen saturation using wearable devices under varying conditions. Aerosp Med Hum Perform. Jan 1, 2023;94(1):42-47. [CrossRef] [Medline]
    48. Hermand E, Coll C, Richalet JP, Lhuissier FJ. Accuracy and reliability of pulse O2 saturation measured by a wrist-worn oximeter. Int J Sports Med. Dec 2021;42(14):1268-1273. [CrossRef] [Medline]
    49. Schiefer LM, Treff G, Treff F, et al. Validity of peripheral oxygen saturation measurements with the Garmin Fēnix® 5X Plus wearable device at 4559 m. Sensors (Basel). Sep 23, 2021;21(19):6363. [CrossRef] [Medline]
    50. Carvalho TD, Pastre CM, de Godoy MF, et al. Fractal correlation property of heart rate variability in chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. Jan 6, 2011;6:23-28. [CrossRef] [Medline]
    51. Stanbury RM, Graham EM. Systemic corticosteroid therapy--side effects and their management. Br J Ophthalmol. Jun 1998;82(6):704-708. [CrossRef] [Medline]
    52. Maclay JD, McAllister DA, Rabinovich R, et al. Systemic elastin degradation in chronic obstructive pulmonary disease. Thorax. Jul 2012;67(7):606-612. [CrossRef] [Medline]
    53. Fine J, Branan KL, Rodriguez AJ, et al. Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring. Biosensors (Basel). Apr 16, 2021;11(4):126. [CrossRef] [Medline]
    54. Rouette J, McDonald EG, Schuster T, Brophy JM, Azoulay L. Treatment and prescribing trends of antihypertensive drugs in 2.7 million UK primary care patients over 31 years: a population-based cohort study. BMJ Open. Jun 10, 2022;12(6):e057510. [CrossRef] [Medline]
    55. Ballin M, Nordström P, Niklasson J, et al. Daily step count and incident diabetes in community-dwelling 70-year-olds: a prospective cohort study. BMC Public Health. Nov 30, 2020;20(1):1830. [CrossRef] [Medline]
    56. Ogliari G, Mahinrad S, Stott DJ, et al. Resting heart rate, heart rate variability and functional decline in old age. CMAJ. Oct 20, 2015;187(15):E442-E449. [CrossRef] [Medline]

    ‎
    ATS: American Thoracic Society
    CAT: COPD Assessment Test
    COMPASS: Composite Autonomic Symptom Score
    COPD: chronic obstructive pulmonary disease
    DAM: Dynaport MoveMonitor
    ERS: European Respiratory Society
    HRV: heart rate variability
    ICC: intraclass correlation coefficient
    mMRC: modified Medical Research Council
    RHR: resting heart rate
    RMSSD: root mean square of successive differences
    RR: respiratory rate
    SpO2: oxygen saturation

    Edited by Lorraine Buis; submitted 03.01.24; peer-reviewed by Chintal Shah, J Edwards, Matthew Fraser; final revised version received 28.06.24; accepted 29.08.24; published 06.11.24.

    Copyright

    © Fien Hermans, Eva Arents, Astrid Blondeel, Wim Janssens, Nina Cardinaels, Patrick Calders, Thierry Troosters, Eric Derom, Heleen Demeyer. Originally published in JMIR mHealth and uHealth (https://mhealth.jmir.org), 6.11.2024.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mHealth and uHealth, is properly cited. The complete bibliographic information, a link to the original publication on https://mhealth.jmir.org/, as well as this copyright and license information must be included.