This study was conducted within the China Healthy Aging Cohort Study (China-Aging; ClinicalTrials.gov identifier: NCT06472674), an ongoing longitudinal, multicenter cohort of community-dwelling older adults across several regions in China [35]. The China-Aging study aims to investigate the prevalence and risk factors of disability and cognitive impairment and to develop predictive models by integrating traditional clinical measures with technology-derived data.

The wearable monitoring component was established within the Beijing center of the China-Aging cohort to collect longitudinal, real-world behavioral and physiological data using wrist-worn wearable devices. The present study reports the feasibility, usability, and acceptability of this monitoring approach under naturalistic conditions. A mixed methods design was used, combining quantitative wearable monitoring data with user experience survey data, including open-ended responses analyzed thematically. Baseline demographic and clinical assessments were conducted by trained researchers as part of the China-Aging cohort procedures.

The study was approved by the Ethics Committee of Xuanwu Hospital, Capital Medical University (approval number 2024214001), and all participants provided written informed consent prior to participation.

Participants were recruited from the Beijing region of the China-Aging cohort through the Baizhifang Community Health Centre. Eligible participants were invited to take part in this wearable monitoring study after completing baseline assessments within the cohort.

Inclusion criteria were (1) age 55 years or above, (2) residence in the local community for at least 1 year, (3) ability to independently use a wrist-worn device, (4) ownership of a compatible Android smartphone for successful device pairing, and (5) ability to install and use the Huawei Health app with regular access to Wi-Fi or mobile data for data synchronization. Exclusion criteria included (1) unstable medical conditions or acute illness at the time of recruitment, (2) severe cervical or lumbar spine disease or advanced osteoarthritis substantially limiting routine daily activity, and (3) incomplete baseline assessment.

Baseline assessments included demographic characteristics (age, sex, education level, marital status, living arrangement, care arrangement, and self-rated health status), cognitive function assessed using the Mini-Mental State Examination (MMSE) [36], and functional status assessed using the Barthel Activities of Daily Living (ADL) Index [37]. All enrolled participants demonstrated sufficient cognitive and functional capacity for independent device use at baseline.

Wearable devices were distributed beginning in December 2023, with rolling enrollment following the completion of baseline assessments. Devices were offered consecutively to eligible and willing participants after the confirmation of smartphone compatibility and successful completion of device pairing, connection testing, and data transmission. Participants were not selected based on demographic characteristics, self-rated health status, or prior technology experience.

Participants received the wearable device at no cost and were instructed to wear it during daily life without restrictions, including during sleep, removing it only when necessary (eg, for charging). The monitoring start date was defined as the day on which participants completed device activation and first wore the device following study onboarding procedures. No predefined monitoring duration or mandatory wear requirements were set, and participants were allowed to discontinue device use at any time without affecting their participation in the China-Aging cohort. Monitoring continued until the last successful data synchronization or the end of November 2025, whichever occurred first.

In-person technical support was provided during initial device setup, with subsequent troubleshooting available through a WeChat group. Brief reminders were sent approximately every 2 days during the first 2 weeks to support device activation and data synchronization and to identify early technical issues. Thereafter, no proactive reminders or adherence-enhancing contact were provided, although participant-initiated technical support remained available through WeChat. This limited-support approach was intended to approximate real-world wearable use under naturalistic conditions.

Participants were provided with a commercial wrist-worn wearable device (Huawei Band 8; Huawei Technologies Co) designed for the continuous monitoring of heart rate, physical activity, and sleep. The device incorporates a PPG sensor for heart rate monitoring and a triaxial accelerometer for activity tracking. Proprietary onboard algorithms generate summary measures of heart rate, step counts, and sleep parameters as implemented by the device manufacturer.

Participants paired the device with the Huawei Health app on their personal smartphones, through which wearable data were transmitted via Bluetooth and synchronized to the Huawei Health Research Platform. Data available through the research platform were pseudonymized and identified by wearable device ID. The linkage between wearable device ID and participant number was maintained separately by the research team. Data exported for analysis were stored on a secure research laptop accessible only to the study team.

Minute-level heart rate data were used as the primary indicator of wear-time, as heart rate is recorded only during direct skin contact [17,38]. Brief data gaps may occur due to transient loss of sensor contact or motion-related artifacts. Consistent with the improved nonwear detection algorithm proposed by Choi et al [39], which extended the foundational approach of Troiano et al [40], short interruptions of 1‐2 minutes were treated as continuous wear, whereas longer gaps were considered potential nonwear periods. In the present study, nonwear was defined as intervals of 90 or above consecutive minutes without valid heart rate data, and subsequent data were treated as a new wear period. This approach was further reviewed and confirmed through consultation with Huawei technical engineers. Accordingly, daily wear-time was calculated as the sum of all valid wear periods within each calendar day.

A valid monitoring day was defined as a day with 10 or more hours of wear time, consistent with definitions used in wearable and accelerometry research [33,41]. A minimum of 3 valid monitoring days has been used in previous wearable research [33]. In the present feasibility study, we additionally required these 3 valid days to be consecutive as a pragmatic operational criterion for identifying short-term monitoring continuity under naturalistic conditions.

Based on these criteria, participants were classified into three monitoring-pattern categories:

User acceptability and usability were assessed using a self-administered questionnaire following a period of real-world wearable use. The questionnaire was developed for this study, informed by existing wearable usability instruments [32,42], and designed to capture multiple dimensions of usability, acceptability, and perceived impact in real-world use. It comprised 16 items, including 13 Likert-scale questions covering device comfort, ease of use, display readability, charging, app connectivity, data recording and accuracy, perceived lifestyle impact, and willingness for continued use and data sharing, as well as 3 open-ended questions exploring perceived benefits, barriers, and suggestions for improvement.

All Likert-scale items were rated on a 5-point scale, with higher scores indicating more positive evaluations. The questionnaire was originally developed in Chinese and translated into English for reporting. It was distributed to all participants after an initial period of device use, and responses were collected for subsequent analysis. Full details of the questionnaire are provided in Multimedia Appendix 1.

All analyses were conducted using R (version 4.5.1) [43]. Descriptive statistics were used to summarize participant characteristics, monitoring metrics (eg, wear-time and valid days), and wearable-derived measures (eg, heart rate, physical activity, and sleep). Continuous variables are reported as mean (SD) for normally distributed data and median (IQR) for nonnormally distributed data.

Group comparisons were performed to examine potential differences in demographic, self-rated health status; cognitive function; and functional characteristics associated with wearable monitoring patterns. The distributions of continuous variables were assessed using the Shapiro-Wilk test. Continuous variables were summarized using median (IQR) and compared across monitoring-pattern groups using the Kruskal-Wallis test, with epsilon-squared (ε²) reported as the effect size. Categorical variables were compared using Pearson chi-square tests or exact tests where appropriate, with Cramér V reported as the effect size. These analyses were exploratory and intended to characterize potential differences associated with monitoring patterns. Statistical significance was defined as a 2-sided P<.05.

For survey data, responses to Likert-scale items were summarized descriptively using median (IQR) and mean (SD). Open-ended responses were analyzed using an inductive thematic approach. Two researchers (YB and YZ) independently reviewed the responses and developed initial thematic categories related to perceived benefits, barriers, and suggested improvements. They then compared their categorizations and discussed disagreements, refining the categories until consensus was reached on the final coding and thematic structure.

To assess the robustness and potential sources of selection bias, we examined alternative monitoring thresholds and continuous monitoring measures, compared baseline characteristics between enrolled and nonenrolled participants, and compared baseline characteristics and monitoring measures between survey respondents and nonrespondents. Details of these analyses are provided in Multimedia Appendix 2.

All 710 participants approached for the wearable study owned a smartphone. Of these, 147 used a phone that was incompatible with the Huawei Health app (eg, iOS devices or unsupported Android versions), 64 did not proceed because stable data transmission could not be established during connection testing, and 13 did not enroll for other reasons that were not systematically recorded. Refusal was not recorded as a separate category and may therefore have been included among these 13 participants. Ultimately, 486 participants received a Huawei Band 8 and entered the wearable study, with wearable data available for all enrolled participants.

Baseline comparisons between the 486 enrolled and 224 nonenrolled participants showed no significant differences in age, sex, education, marital status, or self-rated health. Enrolled participants had higher MMSE scores and differed in living arrangement and care arrangement; a statistically significant but small difference was also observed in ADL. Full results are provided in Multimedia Appendix 2.

Among the 486 enrolled participants, 224 (46.1%) achieved minimal consecutive valid monitoring (≥3 consecutive valid monitoring days), 156 (32.1%) had nonconsecutive valid monitoring, and 106 (21.8%) had no valid monitoring day. Among participants with no valid monitoring day, 97 of 106 (91.5%) participants had a maximum observed daily wear time of less than 1 hour, while the remaining 9 of 106 (8.5%) participants had a maximum daily wear time of 1 to less than 5 hours. Within the nonconsecutive valid monitoring group, 119 of 156 (76.3%) participants had 1 valid monitoring day, 29 of 156 (18.6%) participants had 2 valid monitoring days, and 8 of 156 (5.1%) participants accumulated 3 or more valid monitoring days but never achieved 3 consecutively. Overall, 148 of 156 (94.9%) participants in this group had only 1 or 2 valid monitoring days. Full results are provided in Multimedia Appendix 2.

Baseline demographic, self-rated health status, cognitive, and functional characteristics across the 3 monitoring patterns are summarized in Table 1. The observed age range among enrolled participants was 59 to 86 years. Age, sex, education level, marital status, living arrangement, care arrangement, MMSE score, and Barthel ADL score did not differ significantly across monitoring patterns (all P>.05). Self-rated health was the only measured baseline characteristic that differed significantly across groups (Cramér V=0.114; P=.005), with better self-rated health more frequently reported among participants with minimal consecutive valid monitoring.

Among the 224 participants with minimal consecutive valid monitoring, the median total number of valid monitoring days was 60.5 (IQR 19‐94.3; range 3‐409), while the median maximum consecutive valid-day streak was 16 days (IQR 8‐34; range 3‐168), indicating substantial heterogeneity in both monitoring extent and continuity (Figure 1).

Sensitivity analyses were conducted to examine the robustness of the findings to the primary ≥3-consecutive-valid-day criterion. Among the 380 participants with 1 or more valid monitoring days, 224 (58.9%), 211 (55.5%), 196 (51.6%), and 178 (46.8%) achieved 3 or more, 4 or more, 5 or more, and 7 or more consecutive valid monitoring days, respectively. The association between better self-rated health and consecutive valid monitoring remained directionally consistent across these alternative thresholds. Full results of the sensitivity analyses are provided in Multimedia Appendix 2.

Among participants with minimal consecutive valid monitoring, heart rate and step-count data were available for all 224 participants. Mean daily heart rate was 72.1 (SD 8.6) bpm, and median daily step count was 6177 (IQR 4249‐8872) steps/day. Sleep data were available for 211 of 224 (94.2%) participants, with a median sleep duration of 7.23 (IQR 6.48‐8.04) hours (Table 2).

User acceptability and experience were assessed using a self-administered questionnaire completed by 106 of 486 (21.8%) participants after a period of real-world device use. Survey response rates were higher among participants with minimal consecutive valid monitoring (74/224, 33%) than among those with nonconsecutive valid monitoring (21/156, 13.5%) or no valid monitoring days (11/106, 10.4%). Although respondents and nonrespondents were broadly comparable in measured baseline characteristics (Multimedia Appendix 2), respondents demonstrated greater monitoring continuity and persistence. The survey findings therefore primarily reflect the experiences of participants who maintained greater wearable use.

Across the 13 Likert-scale items (n=106), ratings were generally above the neutral midpoint (score=3), indicating overall favorable usability and acceptability (see Multimedia Appendix 3 for full item-level results). Median scores were 4 for most items, with IQRs ranging from 3-4 to 4-5, and mean values ranging from 3.51 to 4.22, with SDs ranging from 0.59 to 1.09. The highest rating was observed for independence of use (Q6; mean 4.22, SD 0.96), suggesting that most participants were able to operate the device without assistance. Items related to usability and data perception, including comfort, ease of use, readability, charging, connectivity, accuracy, and trust (Q1-Q8), also showed consistently positive ratings.

Perceived impact on health behavior was more varied. While participants reported increased health awareness (Q9) and greater willingness to engage in physical activity (Q10), perceived changes in daily habits (Q11) were more modest (median 3), indicating greater variability in responses. This suggests that although respondents reported greater awareness and motivation, perceived changes in daily habits were more limited. Willingness for long-term use (Q12) and data sharing with health care providers (Q13) remained high (median 4, IQR 4-4), indicating generally favorable acceptability among respondents.

For the open-ended questions, participants reported both perceived benefits and barriers to device use. The most commonly reported benefits included improved sleep monitoring (38/106, 35.8%) and activity or exercise tracking (37/106, 34.9%), followed by increased awareness of heart rate and oxygen saturation (29/106, 27.4%) and greater attention to overall health management (29/106, 27.4%). Only a small proportion (4/106, 3.8%) of participants reported no perceived benefits.

Reported barriers included device discomfort (18/106, 17%), data synchronization issues (12/106, 11.3%), and limited battery life (12/106, 11.3%). Some participants noted that removing the device for charging or showering disrupted routines and increased the likelihood of forgetting to put it back on. Others reported difficulty using functions beyond basic activity tracking. In contrast, 36 of 106 (34%) participants reported no inconvenience and were satisfied with device use.

Suggestions for improvement focused on both functionality and usability. The most common suggestions included adding additional health monitoring features (35/106, 33%), such as blood pressure or glucose tracking, simplifying device operation and guidance (17/106, 16%), and improving device comfort and battery performance (12/106, 11.3%). Approximately one quarter (27/106, 25.5%) of respondents reported no additional suggestions and were satisfied with the current device. A detailed breakdown of thematic categories is provided in Multimedia Appendix 3.

This study evaluated the real-world feasibility, usability, and acceptability of using a consumer wearable device for continuous health monitoring in community-dwelling older adults under limited-support, naturalistic conditions. The findings indicate that wearable monitoring was feasible for a substantial proportion of participants, with many achieving consecutive valid monitoring and maintaining device use beyond the initial monitoring period. At the same time, monitoring patterns varied considerably across individuals, with substantial heterogeneity in both monitoring extent and continuity, highlighting challenges for sustained real-world monitoring and large-scale implementation.

The observed monitoring patterns should be interpreted in the context of the limited-support design. Participants received support during initial device setup and brief reminders during the first 2 weeks but no proactive adherence support thereafter. This differs from previous studies using structured follow-up, training, prompting, or incentives and reporting relatively high adherence [16,32,44,45]. For example, Paolillo et al [45] reported wearable use on an average of 95% of study days during a supported 30-day protocol, while Holmqvist et al [44] reported a mean daily wear time of 21 hours during a structured 4-week study involving training and automated prompting. Although direct comparisons are limited by differences in study design and monitoring definitions, the present findings provide an estimate of wearable use under conditions with limited ongoing support. The substantial variation in monitoring extent and continuity also aligns with broader patterns of digital health use, in which sustained engagement depends on the integration of technology into everyday routines [46,47]. Together, these findings suggest that wearable feasibility is influenced by implementation context and support intensity and that supported and naturalistic studies provide complementary evidence for understanding real-world use.

From an implementation perspective, participants with nonconsecutive valid monitoring may represent an important target for improving sustained wearable use. Most participants in this group accumulated only 1 or 2 valid monitoring days, suggesting that initial device use did not consistently develop into sustained monitoring [48,49]. Practical barriers identified in the user experience survey, including data synchronization, battery management, and interruptions associated with device removal, may contribute to difficulties in incorporating wearable use into everyday routines [29,32,50]. Previous research suggests that sustained wearable use can be supported by the development of routine behaviors, such as integrating device wear and charging into established daily activities [50,51]. Additional support during the early monitoring period may therefore be one strategy for facilitating the transition from initial to more sustained use.

Participants with no valid monitoring day may represent a different implementation challenge. Although all participants in this group generated some wearable data, observed wear was extremely limited: among the 106 participants, 97 (91.5%) participants had a maximum daily wear time of less than 1 hour, and none reached 5 hours on any observed day. This pattern suggests that the predominant challenge was establishing meaningful wearable use after initial setup, rather than wear that narrowly fell below the ≥10-hour valid-day criterion. Notably, this occurred despite successful device pairing, connection testing, and data transmission before enrollment, suggesting that successful technical onboarding alone may not be sufficient to ensure subsequent wearable use. Potential barriers may include technical and usability challenges, perceived usefulness, and difficulties integrating the device into everyday activities [29,32,52,53]. However, the specific reasons for limited monitoring in this group could not be determined because survey participation was particularly low, and barriers reported by survey respondents may not reflect the experiences of participants with little valid monitoring. Further research is therefore needed to identify the factors contributing to limited early-stage monitoring and determine whether targeted technical or usability support could facilitate continued device use.

Taken together, these findings suggest that different monitoring patterns may require different implementation strategies. Participants with some valid monitoring but limited continuity may benefit from support focused on establishing consistent monitoring routines, whereas those who do not achieve a valid monitoring day may require greater attention to barriers occurring earlier in the monitoring process. Distinguishing between these patterns may therefore help inform more targeted approaches to supporting wearable monitoring.

Self-rated health was the only measured baseline characteristic significantly associated with monitoring pattern, with better self-rated health more frequently reported among participants achieving minimal consecutive valid monitoring. This association remained directionally consistent across alternative consecutive-day thresholds, suggesting that the finding was not dependent on the primary ≥3-day criterion. Similar associations between health status and digital health use have been reported previously, with greater health burden potentially increasing the practical demands of device use [34,48,49]. Self-rated health is a multidimensional measure that may reflect aspects of physical and psychological well-being, perceived control, and health-related self-efficacy, which have also been associated with technology adoption and health behaviors [52,54]. However, the mechanisms underlying the observed association were not assessed in the present study, and the finding should therefore be interpreted as exploratory rather than causal.

The absence of significant associations between monitoring patterns and age, education, cognitive function, or functional status is also informative, suggesting that these measured characteristics did not substantially differentiate monitoring patterns within the enrolled cohort. However, these findings should be interpreted in the context of the relatively cognitively and functionally intact study population. The high and narrow range of MMSE and Barthel ADL scores limited the ability to examine the effects of greater cognitive or functional impairment. In addition, comparisons between enrolled and nonenrolled individuals identified differences in cognitive function, living arrangement, and care arrangement, indicating that selection may occur before wearable monitoring begins. Therefore, the absence of associations within the enrolled cohort should not be interpreted as evidence that cognitive or functional characteristics are not important for wearable adoption more broadly [52,53,55]. Further research including older adults with more diverse cognitive, functional, and digital capabilities is needed.

Among participants with minimal consecutive valid monitoring, wearable-derived heart rate and step-count data were available for all participants, while sleep data were available for 94.2% (n=211) of the participants. This indicates that multiple behavioral and physiological measures could be obtained among participants who achieved minimal consecutive valid monitoring. The slightly lower availability of sleep data may reflect the additional requirements of overnight wear, including charging and device removal. More broadly, these findings indicate that successful wearable monitoring does not necessarily result in equivalent data availability across monitoring domains, an important consideration when consumer wearables are used for longitudinal health monitoring.

The user experience survey also provided insight into practical aspects of device use. While some participants reported no inconvenience, others identified barriers related to comfort, data synchronization, and battery life, and 33% (35/106) of the respondents suggested additional health-monitoring functions, such as blood pressure or glucose monitoring. These findings suggest that both practical burden and perceived relevance may influence continued wearable use and that greater alignment between device functionality and users’ health priorities may enhance acceptability [29,32,46,47]. However, these findings should be interpreted with caution because the respondents demonstrated greater monitoring continuity and persistence than nonrespondents, indicating that the usability, acceptability, and barrier findings disproportionately reflect the experiences of participants who maintained greater wearable use.

The principal strength of this study lies in its community-based design under naturalistic conditions, which captured variability in wearable use with limited ongoing support. The combination of objective monitoring data and user experience responses also enabled feasibility to be examined from complementary perspectives. Several limitations should be considered. First, recruitment from an urban community setting and requirements for smartphone compatibility and successful data transmission may limit generalizability, particularly to older adults with lower digital capability or greater cognitive or functional impairment. Differences between enrolled and nonenrolled participants further indicate potential preenrollment selection. Second, the ≥3-consecutive-valid-day criterion was a pragmatic operational definition rather than a validated threshold, although sensitivity analyses using alternative thresholds and continuous monitoring measures supported the overall findings. Third, the relatively low user experience survey participation rate (106/486, 21.8%) and greater monitoring persistence among respondents may have resulted in more favorable estimates of usability and acceptability. In addition, individual-level use of technical support was not systematically recorded, limiting our ability to assess whether support use differed across monitoring patterns or influenced monitoring continuity. Finally, the study used a single consumer wearable ecosystem with proprietary algorithms that were not independently validated against research-grade reference measures in this population. Future research should therefore examine longer-term monitoring patterns, identify barriers to sustained use, evaluate targeted technical and behavioral support, and validate wearable-derived measures against reference standards. Longitudinal associations with cognitive, functional, and health outcomes should also be investigated to establish the potential clinical relevance of these measures.

In conclusion, this study demonstrates that wrist-worn wearable monitoring is feasible for a substantial proportion of community-dwelling older adults in China under limited-support, naturalistic conditions, although monitoring extent and continuity varied considerably. Usability and acceptability were generally favorable among survey respondents. Future research should identify strategies to support sustained monitoring and establish the clinical relevance of wearable-derived measures for longitudinal changes in cognitive and functional health.