This scoping review was guided by the framework and principles reported by Arksey and O’Malley [34] and in accordance with the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guideline [35]. A scoping review provides a literature overview by mapping key concepts in the evidence base of the research field, which can be used to inform needs and identify knowledge gaps [36]. PRISMA-ScR checklist for this review is presented in Multimedia Appendix 1.

A comprehensive search strategy was developed by information and health specialists using a combination of MeSH (Medical Subject Headings) terms and free-text terms. The databases searched were PubMed, Web of Science, CINAHL, Association for Computing Machinery (ACM), IEEE Xplore, and ScienceDirect from inception to July 28, 2024. In addition, relevant studies were identified through a manual search of the reference lists from all available records and conference proceedings obtained in the initial search. The actual strategies listing all search terms used and how they are combined are available in the Multimedia Appendix 2.

Eligibility for our scoping review was determined using the PCC (Population, Concept, and Context) framework, which is listed in Textbox 1.

We imported the retrieved records into EndNote X9 (Thomson Reuters Scientific LLC) for management. The full texts of the included articles were independently evaluated by 2 reviewers. Disagreements regarding selection and inclusion were resolved through consensus-based decision-making.

The data extraction form was based on a previous publication and adapted to the needs of this review on the same topic, and we randomly selected 10 studies to test and refine it. The included studies obtained information on target population, sample size, age, wearable strategies, measurements, main results, and conclusions. The data extraction process was performed by two authors, and inconsistent data were resolved by the third author.

Data extracted from primary studies were mapped to visually summarize outcome measures identified and coded. We summarized the results using a narrative descriptive synthesizing approach and presented them in tables and figures. The evidence map presents outcome measures by tabulating the type of wearable devices against the domain of the outcome measure. The data extraction form used in this review was piloted and is shown in Multimedia Appendix 3.

The schematic flow for the selection of the studies included is shown in Figure 1. In total, 3798 studies were identified. After removing duplicate citations, 3076 studies remained for the title and abstract screening, of which 222 were considered potentially eligible studies for full-text review. Among them, 109 papers met the inclusion criteria after full-text review [37-125].

The number of studies generally increased over time, except for a decline in 2018 (Table 1). They were conducted across 26 countries, with more than half of the studies originating from European countries and the United States (Figure 2 and Multimedia Appendix 4). Most of the included participants mostly did not report any health conditions (51/109, 46.79%), The remaining studies primarily focused on populations with frailty, Parkinson disease, atrial fibrillation, chronic obstructive pulmonary disease (COPD), stroke, and other conditions. The study designs were diverse, the application design studies (38/109, 34.86%) and observational studies (37/109, 33.94%) formed the largest proportion, accounting for nearly two-third of all studies. The characteristics of the included studies are presented in Multimedia Appendix 4.

The most commonly reported health targets of wearable devices were (1) mobility (n=23/109), (2) mental health (6/109), (3)fall-related (18/109), (4) arrhythmia detection (10/109), (5)activity recognition (22/109), and (6) disease diagnosis (6/109),and (7) sleep monitoring (8/109) and other conditions. The results are shown in Figure 3.

In this study, we found two types of wearable devices evaluation approaches: (1) metrics for performance evaluation in a controlled laboratory and clinical setting and (2) evaluations involving end users in real-world settings. Analysis revealed that different studies use a variety of metrics for performance evaluation. Despite variations in machine learning algorithms, the figure showed the evaluation metrics used across 45 studies. In our study, 39.45% (43/109) of included studies used the metrics to assess performance. Sensitivity (29/109, 26.61%), accuracy (23/109, 21.10%), specificity (22/109, 20.18%), precision recall (11/109, 10.09%), and F₁-score (11/109, 10.09%) were the most used metrics among offline evaluation metrics. Other popular offline evaluation metrics are accuracy-related measurements, such as root-mean-square error (3/109, 2.75%), and the mean absolute error (1/109, 0.92%).

In addition to performance testing, there were also 34 studies involving end users that evaluated the effectiveness, safety, user acceptability, usability, adherence, satisfaction, user experience, cost, and satisfaction. Clinical effectiveness (n=22, 20.18%) and acceptability (n=17, 15.60%) were the most commonly evaluated domains. Specifically, only one study (n=1, 0.92%) evaluated the user experience and satisfaction. The details are illustrated in Multimedia Appendix 4.

There were several challenges and opportunities associated with wearable devices that were reported in studies or inductively emerged based on study descriptions (Table 4 and Figure 5). The opportunities presented by wearable devices are countered by multiple challenges, including data availability and reliability, technical limitations, utility and user acceptance, cost, security and privacy, performance gaps, and challenges.

Our study is the first to conduct a systematic scoping review and create an evidence map to provide an overview of the state of evidence regarding the development and evaluation of wearable devices. A total of 109 studies on wearable devices were included in our systematic review. This study provided a detailed mapping of existing wearable devices, outlining their basic characteristics, health targets, and underlying technologies, while also showing the evaluating results and identifying the challenges and opportunities associated with their use by older adults.

Mobility, fall-related incidents, and mental health monitoring are the primary focuses of wearable devices designed for older adults. Mobility-focused devices often include features such as step counting, gait analysis, and balance tracking to help users maintain their physical activity levels, promote healthy behaviors like exercise, and offer a wide range of applications in managing various diseases, such as high-risk older patients in the preoperative stage [126], hospitalized older people [127], patients with cancer [128], those with Parkinson disease [129], and those with COPD [130]. Faller classification in older adults can facilitate preventative care before a fall occurs. While mobility is prominent, mental health and cognition also remain important considerations for wearable technology targeting older adults. Early identification and monitoring of mental illnesses may be quite challenging. Contrary to the results of traditional patient self-reports, wearable devices could assess psychological states by continuously monitoring other indicators. Chan et al [131] used wrist-worn accelerometers to investigate whether gait biomarkers could predict the incidence of depressive episodes in older individuals and indeed found that these biomarkers can serve as predictive indicators. Toosizadeh et al [132] confirmed significant associations between upper-extremity speed, range of motion, and speed variability with both the Montreal Cognitive Assessment (MoCA) score and the gait performance within the dual-task condition. Vital signs monitoring is a major component of wearable devices, primarily including heart rate. The detection and diagnosis of atrial fibrillation (AF) are often challenging due to the asymptomatic and intermittent nature of AF. For patients who may be at risk of cardiac arrest, wearable devices can help monitor arrhythmias. The novel Necklace-ECG device is able to detect AF obtained with novel Necklace-ECG device [132], mobile device [133,134]. Santala et al [135] have demonstrated that an all-time wearable Necklace-ECG can detect and diagnose AF with high accuracy. In addition, there were measurements of physiological indicators, such as sleep calmness. Abbas et al [136] use devices that monitor daily steps, sleep duration, and quality, as well as activity levels, to distinguish physical frailty from daily routine. In summary, existing evidence on the broad implementation of wearable devices in clinical care highlights their benefits, which include enabling patient-centered care, enhancing the monitoring of physical activity, facilitating timely treatment adjustments, reducing the occurrence of unexpected events, and minimizing the need for unnecessary clinic visits.

Our systematic review explores how wearable devices are evaluated. Comparing sensor technologies with clinical scales and performance measures in a controlled laboratory and clinical setting is crucial for evaluating device effectiveness in specific tasks. The performance of the algorithm depends on several factors, including the dataset used, the type of classifier, the data acquisition device, and the ratio of training to testing data [137]. Performance testing of wearable devices typically evaluates key metrics such as accuracy (data and classification), reliability (stability and failure rates), precision and recall, sensitivity and specificity, and F₁-score. Researchers have noticed this lack of user participation in single wearable real-world studies, and it has been identified as the main challenge in this field [138]. Although there has been an increasing trend in studies focusing on using wearable devices for predicting or quantifying the response to an intervention. In addition to effectiveness, only a very small number of studies have investigated the acceptability of the technology, user adherence, user experience, device cost, and safety. For example, Ward et al [130] reported on findings from a single-arm trial that an activity monitor technology could be able to support effective remote walking exercise prescription and participation during pulmonary rehabilitation in COPD. Preliminary findings from 3 studies [139-141], which assessed safety by reporting on related adverse events, indicated that the intervention was safe and well-accepted by the participants. Conducting a cost-benefit comparison between programs implemented in health care systems is challenging, and only one study explored the costs of home pedometer-assisted physical activity in patients with COPD [142]. The Model for Assessment of Telemedicine Applications (MATA) offers a structured framework for the multidisciplinary evaluation of telemedicine applications. This model assists decision makers in selecting the most efficient and cost-effective technologies, taking into account the specific characteristics of the application, safety, clinical efficacy, patient perspectives, and economic, organizational, sociocultural, ethical, and legal considerations [143]. Due to the limitations of laboratory studies in terms of their observation window (min to h) and the environment (generally safe, no obstacles) [144,145]. Researchers and developers can gain a deeper understanding of the real-world impact and effectiveness of wearable devices in health monitoring [146], ultimately leading to better-designed and more impactful solutions for end users. Wearable technology may be able to support effective remote walking.

There are many barriers to the full clinical adoption of digital monitoring, including technical limitations that hinder usability and satisfaction, as well as the lack of clear guidelines for wearable applications and data collection. First, the technical limitations of wearable devices, including interoperability issues, short battery life, large device size, lack of personalized settings, and inadequate support, restrict their application scenarios and may lead to a less satisfactory user experience by not catering to individual preferences [147,148]. Given the limitations observed in the structural design and practical applications of wearable technology, we can undertake several optimizations to improve their effectiveness in clinical practice. Second, our review demonstrated that there is substantial heterogeneity among the wearable sensor technologies used for the assessment of older adults. A large proportion of the devices, their potential benefits, and utility are still far from implementation in clinical routine [149]. Besides, future research should also focus on identifying factors that could enhance the effectiveness of wearable technologies and promote their acceptance and adherence in the longer term. Such factors include the digital technology types (eg, user-friendly design and personalized settings) and the nondigital elements in digital health interventions (eg, motivation and engagement) [25]. Third, wearable devices face major challenges including inadequate data availability and reliability, as well as privacy protection concerns [150]. Food and Drug Administration [151] and the European Medicines Agency [152] have developed guidelines for the use, validation, and data reporting of wearable sensor technology in clinical trials. However, these guidelines are not yet widely known, and journals should recommend and even require their adoption to disseminate these best practices [153,154]. Last, given the complexity of human responses and the variability among individuals, data collection can be quite challenging [155]. However, the data collection process is often both costly and time-consuming, leading most academic studies to rely on relatively limited datasets. To address this, it is advisable to gather as much data as possible from participants in order to develop robust algorithms for data cleansing and to construct more accurate and broadly applicable models. Adopting a transparent and reproducible approach to data collection, training, and model evaluation is highly recommended to bolster confidence in research outcomes and to facilitate the effective accumulation of research efforts [156].

Despite many barriers to their application, wearable devices have brought huge benefits to the aging global population. In addition, their impact on health care is profound as the younger, technologically literate generation ages into adulthood. On the one hand, the potential to significantly boost the technological literacy of future older populations becomes more apparent. Currently, older adults may have faced a double exclusion due to a lack of digital literacy and social interaction who rely on health monitoring technologies for personalized medical guidance [157]. In contrast, future cohorts of older adults, being more adept with data and platforms, may actively better understand the data and recommendations provided by the devices, leading to more active participation in remote health education programs. On the other hand, the dynamic and continuous accumulation of health data contrasts sharply with traditional, static, and episodic medical history records. This shift has the potential to revolutionize the entire health care system, moving health care from a “diagnosis-treatment model” toward a “prediction-prevention model.” For instance, existing research indicates that early signs of cognitive function may be identified through sleep data [158]. Consequently, future medical services may place greater emphasis on the prediction and prevention of disease, potentially reducing the need for treatment approaches.

Furthermore, improvements are needed for (1) research should include a broader and more diverse range of older adults to ensure that the findings are applicable to various demographic groups, which includes considering differences in age, gender, socioeconomic status, and cultural backgrounds, (2) understanding the underlying physiological processes that these technologies monitor, and influence can lead to more effective and targeted interventions. This includes studying how sensors and algorithms interact with the body’s biological signals to provide accurate and reliable data, (3) research should extend beyond laboratory settings to include various clinical and care environments. This will help identify the practical challenges and benefits of using wearable technologies in real-world scenarios, such as hospitals, nursing homes, and home care settings, (4) outcome evaluations are necessary to identify the factors that can enhance the health benefits of digital technologies. This includes assessing both short- and long-term outcomes, with a focus on health technology assessments, including the effectiveness, patient safety, and impact on health care costs, (5) Technologically literate older adults may be more proactive in managing their health, leveraging wearables for continuous monitoring of vital signs, chronic conditions, and mental health. Adherence could be further enhanced with the provision of additional cognitive aids, such as detailed documentation of study procedures and comprehensive user manuals, as well as by allowing sufficient time for users to establish wearing routines [11], (6) by integrating quantitative and qualitative insights from participants, we can further refine wearable device technology. The research about how to simplify interfaces and operations to make them more intuitive for older adults, while exploring ways to tailor devices based on individual health conditions, lifestyle, and preferences to provide personalized health management and reminder services.

First, not all research involving wearable devices was identified by our search. For example, some company products may not appear in databases. Due to practical constraints, we included only studies published in the English language. Consequently, it is likely that we missed some studies published in other languages, potentially leading to an incomplete and potentially biased set of results. Second, we reported on the challenges and opportunities of wearable devices from the included studies. While these insights may be helpful, they are not objective measures. Last, the categories of research and the objectives for using wearable devices introduced in our study may overlap, as separation and classification are artificial.

We see a growing uptake of wearable technology in health research with a notable trend toward its use in older adults. In the future, a broader and more diverse range of studies should be designed to investigate which types of wearable devices are most beneficial for older adults. In addition, it is important to explore the potential of emerging innovations in enhancing health outcomes. Focusing on optimizing wearable device technology to improve usability, and effectiveness is also crucial. Outcome evaluations are needed to identify factors that could enhance the health benefits of these technologies and to measure their real-world impact. This review can help nonmedical professionals and policy makers visualize and better understand wearable technology in future studies.