Technology Design (n = 121)

Technology design studies involved the development and assessment of wearable technologies. The three related themes were (a) feasibility (n = 89), which comprised proof-of-concept studies that tested novel devices, technological systems, and software; (b) validity (n = 31), which comprised studies validating the accuracy of the measurements from wearables in specific conditions and populations (these typically compared the measurements from wearables with gold-standard measurements from established medical equipment used in clinical settings); and (c) usability (n = 5), which entailed studies that explored how safe, comfortable, and easy to use wearables were for users.

We noticed that most wearable technologies were in the initial testing phase (feasibility and validity), while the usability studies were comparably limited. Much of the input research was conducted within the laboratory (n = 66) rather than in commercial settings (n = 55) with general consumers or in clinical settings with patients. The focus of input factor is on technology-centered characteristics rather than on end-user adoption characteristics.

Feasibility studies (n = 89) explored the use of wearables for mental health (n = 22) and physical health (n = 67). Over time, wearables are measuring mental health–related information more precisely. An initial study only discriminated stress from normal workload (Setz et al., 2010). Subsequently, measurements became more sophisticated, being able to distinguish between four types of mental state (depression, mixed state, hypomania, and euthymia) in persons with bipolar disorder (Valenza et al., 2014), then proceeding to measure continuous anxiety levels (Betti et al., 2018) and distinct types of emotional states (neutral, tenderness, amusement, anger, disgust, fear, and sadness; Martínez‐Rodrigo et al., 2020).

Wearables for physical health have similarly progressed in sophistication over the years. Initially, wearable activity trackers solely provided binary information of active or sedentary behavior (Kelly et al., 2011). Later, with integrated heart rate sensors and accelerometers, wearables for monitoring aerobic activities were developed (Reiss & Stricker, 2014). In addition to monitoring the amount of activity, the quality of activity could be monitored, for instance, gait quality (Schwenk et al., 2015). Further, the performance of wearables for the classification of different activity types has been improved by incorporating signals other than motion, such as temperature (Lui & Menon, 2019).

In addition to directly measuring mental states and physical activity, studies have suggested the feasibility of wearables for reflecting high-level constructs of health-related status. For example, recent studies have proposed using calculations based on activity intensity and duration for predicting physical fatigue (Sedighi Maman et al., 2017) and using gait quality to reflect cognitive ability, thus supporting the diagnosis of dementia (Thorpe et al., 2019).

Likewise, validity studies (n = 31) tested novel wearables for measuring signals reflecting mental health (n = 4) and physical health (n = 27). Those focused on mental health tested the accuracy of the E4 wrist band (Menghini et al., 2019), Apple Watch 3 (Hernando et al., 2018), Fitbit Charge 2 (B. W. Nelson & Allen, 2019), and an integrated system (based on electrocardiogram, electrodermal activity, and electroencephalogram; Betti et al., 2018) for assessing mental status. These studies offered optimistic conclusions about wrist-mounted devices providing accurate estimations of stress in laboratory tests.

For physical health monitoring, validity examinations assessed the accuracy of wearables for measuring step counts (n = 13), sleep (n = 7), gait (n = 9), and electrocardiogram and respiratory signals (n = 1). Overall, the results were less conclusive. First, wearables for step counting produced valid measurements when the research population comprised healthy young people (Ferguson et al., 2015), but when the target population constituted patients with chronic diseases, a significant underestimation of steps was reported (Ummels et al., 2018; Wong et al., 2018). According to the results of validation tests, relatively large measurement errors appeared when wearables were used for measuring activity energy expenditure (Ferguson et al., 2015), walking distance (Compagnat et al., 2019), and activity intensity (Byun et al., 2018). Second, sleep measurement studies showed that consumer-level wearables had high accuracy in measuring total sleep time, but encountered limitations in monitoring deep sleep (de Zambotti et al., 2018) and performed poorly in detecting wake-up times after onset of sleep (J.-M. Lee et al., 2018). Third, studies showed that clinical-grade wearables provided a valid measurement of gait parameters (Fusca et al., 2018) and could capture fall risks with accuracy (Di Rosa et al., 2017). These studies on gait suggest that wearables show promise in the prevention of potential injury in high-risk populations such as older people with chronic diseases and those suffering from obesity. Fourth, one study designed an algorithm based on the unsupervised isolation forest model to reduce false alarms of arrhythmia (Xu et al., 2021).

Last, usability studies (n = 5) assessed how well users responded to the design of wearables. For instance, testing social communication coaching smart-glasses with users who had autism spectrum disorder suggested usability for a range of people of diverse ages and condition severity (Keshav et al., 2017). In another study, researchers confirmed that skin integrity was maintained in older adults who wore sensors for fall detection (Ferrari et al., 2012). Three other studies applied the System Usability Scale evaluation, with two suggesting that a wearable soft-robotic glove received high usability scores in supporting hand functions (Radder et al., 2018), and a smart work cloth proved usable as a workload risk assessment tool (Yang et al., 2018). The final study found that the usability scores of seven mainstream wearable devices were positively correlated with activity type identification function and negatively correlated with price and expandable new features (Liang et al., 2018).

In general, studies reported that users responded well to the physical forms of wearables. At the same time, usability varies between different wearables due to variation in human–computer interface design and other factors such as functions and financial costs.

User Acceptance (n = 61)

Studies on user acceptance involved agreeing with and approving wearables from the end-users’ perspective. We noted two themes in this category: (a) attitudes (n = 55), from the viewpoints of either potential or actual users toward wearables; and (b) usage (n = 17), which comprises statistics of users’ usage patterns. Most studies investigated attitudes, while only a small number of studies involved usage surveys.

Studies on attitude (n = 55) aimed to collect subjective user perceptions toward novel wearable products. According to surveys, wearable technologies were perceived as fun to use for some people (Naslund et al., 2016) and were also considered useful in terms of providing users with increased knowledge (Jiwani et al., 2021).

However, there were also negative attitudes toward wearables. The most frequently noted concern was the risk of privacy violation (Greenfield et al., 2016). Other concerns were related to stability and maturity of the technology, such as charging, synchronizing, compatibility, data accuracy (Grym et al., 2019), functional limitations (Ehn et al., 2018), and inopportune alerts and attention (Leonard et al., 2017). In addition, specific concerns existed among some populations. For instance, older people with technical anxiety often perceived wearables as difficult to use (Fang & Chang, 2016). Some employees were wary of the massive data trail that businesses could harness and repurpose for other goals (Mettler & Wulf, 2019). Clinicians had concerns about insufficient technical support, work overload, and the risk of patients becoming addicted to data regarding wearables in clinical settings (Bellicha et al., 2017).

Usage (n = 17) surveys found that the actual use behavior of wearables varied by consumer demographics and socioeconomic status. A survey in Australia showed that the use of trackers was lower in male, older, less-educated, nonworking, and inactive participants (Alley et al., 2016). A Canadian survey yielded similar results, finding that users of digital self-trackers were typically young, healthy, employed, university-educated, and with a higher annual family income (Paré et al., 2018). Regarding usage patterns, an exponential decay was observed in a study conducted in France; of 711 people receiving a Fitbit device, 562 were still tracking after 100 days, and 114 tracking after 320 days (Hermsen et al., 2017). One study compared the characteristics of current and former users and found that former users had a lower body mass index, reported fewer medical conditions, shared data from their device less often, and received the device as a gift more frequently (Friel et al., 2020).

Engagement Strategy (n = 24)

Studies on engagement strategies investigated user-centered designs based on psychosocial factors to foster acceptance and adherence of wearable-based health interventions. We identified four types of engagement strategies: (a) social incentives (n = 11) leveraged the influence of social networks involving experts, friends, family, and so on; (b) information feedback (n = 10) included content provided to users about their health status, goal achievements, or performance benchmarking; (c) financial incentives (n = 4) were monetary rewards used in the combination of wearable health interventions; (d) and gamification (n = 3) was the incorporation of game elements such as competition, points, and virtual gifts to improve participation, loyalty, and fun.

Social incentives (n = 11) were informed by social support or social comparison theories. Social support included emotional, informational, instrumental, or companionship resources mediated by technology that one could get from peers, family members, experts, and other people in their social network. Studies suggest that social support, in the form of motivational interviewing (Aschbrenner et al., 2016), habit education (Ellingson et al., 2019), and consultations (Lyons et al., 2017), have the potential to enhance the effect of wearables for increasing physical activity. Social comparison theory suggests that such benchmarking may provide an inspiration to improve; for example, participants who received messages that compared their calorie-burning exercise with that of a friend were more willing to perform preventive healthcare measures (D.-H. Shin & Biocca, 2017).

In general, wearable interventions, in combination with social incentives, helped to improve self-efficacy (Colón-Semenza et al., 2018), enhance motivation (Rönkkö, 2018), and reinforce habit formation (Ellingson et al., 2019). The additional benefits of social incentives were examined in only one study, which found no significant differences (a small increase of ~48 steps/day) between groups that used Fitbit individually or in combination with health coaching (Ellingson et al., 2019).

Information feedback (n = 10) as an engagement strategy is informed by conceptualization from human–computer interaction theories to support self-monitoring for achieving health goals. Various formats of information feedback were designed, and their effects were compared. In one study, intelligent auditory cues performed better than verbal feedback in improving gait (Ginis et al., 2017). In another, messages using egocentric statements and contextual goal-setting were more useful than allocentric and fixed-goal messages for health behavior change (Jang & Kim, 2019). Researchers found that text walking distance was more effective than image walking routes in encouraging physical activity participation (D.-H. Shin & Biocca, 2017). Participants with bite-count feedback combined with a bite-count goal consumed less food (Jasper et al., 2016), while motion feedback encouraged participants to stretch more (S. Kim et al., 2018). Idle alerts (Lyons et al., 2017) and activity reminders (Wang et al., 2015) based on wearables were useful for increasing physical activity.

Information feedback helped support self-monitoring, goal-setting, prompts, and cues, and therefore encouraged individuals to participate in healthy behaviors. One study indicated that when wearable-based interventions were combined with short message service (SMS) to prompt physical activity, the intervention group increased their activity by 1,266 steps/week, compared to the control group, which solely used wearables, although the SMS effect lasted for only a week (Wang et al., 2015).

Financial incentives (n = 4) are designed based on various psychological principles to enhance the motivation of individuals. Regarding physical activity, studies suggested that a loss-framed financial incentive was effective (Chokshi et al., 2018). Alternatively, a tiered incentive design was more effective compared with a fixed amount (Norman et al., 2016), while rewards in cash were more effective than in the form of a charity donation in the name of participants (Finkelstein et al., 2016). The addition of financial incentives to wearable-based interventions, with total possible incentive of US$268, induced an additional 2-kilogram weight loss in a 12-week intervention (D. W. Shin et al., 2017).

Financial incentives were reported as effective in all four studies, but it was unclear whether the effect could be sustained after the incentives were removed. Among the four studies, two did not investigate the follow-up period, one found that activity levels fell back to baseline values, and another claimed the effects were sustained.

Gamification (n = 3) was applied to promote healthy behavior through game-based persuasion. Studies awarded virtual medals (S. Kim et al., 2018) and trophies (Ratz et al., 2019) as rewards for exercise goal attainment. An online game (Clans of Oulu) based on interaction between users allowed players to use earned points to claim area for their clan (Leinonen et al., 2017).

In terms of the effectiveness of gamification, Kim et al. (2018) indicated that the simple design of virtual medal rewards did not show effectiveness, whereas Clans of Oulu, which included enriched game elements such as competition, conflict, collaboration, and scoring or rewards, was effective in promoting moderate-to-vigorous activity (Leinonen et al., 2017).

Adoption (n = 22)

In studies on adoption, the TAM (n = 9) and the unified theory of acceptance and use of technology (UTAUT/UTAUT2; n = 8) were the most commonly used models to understand the adoption of wearables.

Perceived usefulness and perceived ease of use are the two main underlying constructs in TAM (Davis, 1989). Studies confirmed that perceived usefulness has a significant impact on consumer intention to adopt wearables, while the reported path coefficient varied broadly, from r = 0.21 (Choi et al., 2017) to r = 0.69 (J. Li et al., 2019). Perceived ease of use appeared less important in explaining the adoption of wearables. Some studies showed that the path coefficient was either insignificant (J. Li et al., 2019) or relatively small (the largest path coefficient reported was r = 0.15; Choi et al., 2017).

The UTAUT unified several previous models, including TAM, and identified four determinants—that is, performance expectancy, effort expectancy, social influence, and facilitating conditions (Venkatesh et al., 2003). UTAUT2 extended the model by integrating hedonic motivation, price, and habit (Venkatesh et al., 2012). Existing studies reported mixed findings about the determinants of users’ intentions to adopt wearables. Some studies reported that performance expectancy and effort expectancy positively affected adoption intention (Wang et al., 2020). However, one study found the two path correlations significant for the adoption of wearable medical devices, but insignificant for the adoption of wearable fitness devices (Gao et al., 2015).

Across the adoption studies, factors concerning the unique characteristics of wearable technologies that directly or indirectly influenced adoption were compatibility (J. Li et al., 2019), usability (Rupp et al., 2018), reliability (J. Kim, 2014), functional congruence (Gao et al., 2015), interactivity (Park et al., 2016), perceived informativeness (H. Li et al., 2016), perceived irreplaceability (Zhang et al., 2017), and perceived credibility (Zhang et al., 2017). At the same time, the influence of incomplete data (Hardy et al., 2018) was not supported. A comparison of the effect size of those factors was not possible since these rarely appeared in studies with high heterogeneity.

Individual differences also explained adoption. Studies revealed that age (Rupp et al., 2018) and self-reported health conditions (J. Li et al., 2019) have negative effects on adoption intention. Self-efficacy, physical activity level, personality (Rupp et al., 2018), innovativeness (Macdonald et al., 2019), and health interest (S. Y. Lee & Lee, 2018) have positive impacts on adoption intention. However, the effect of experience with the technology was not supported (Choi et al., 2017).

Theory-based adoption models provide understanding of the underlying psychosocial mechanisms of how technical features, individuals’ internal psychological factors, and external environmental factors impact adoption. Self-determination theory (SDT) and privacy calculus theory (PCT) were applied in the development of adoption models. In line with SDT, consumers tended to adopt wearables with greater affordances to support self-determination needs—namely, autonomy, competence, and relatedness (Rupp et al., 2018). In line with PCT, people’s decisions to adopt wearables are determined by risk-benefit analysis. If perceived benefits are higher than perceived risk, then they are more likely to adopt the wearable device (H. Li et al., 2016).

Research models based on technical-centered theories (i.e., TAM, UTAUT/UTAUT2) showed strong power in explaining individual adoption intention toward wearable products, with reported R² values ranging from 32% (Zhang et al., 2017) to 69% (J. Li et al., 2019). Models based on PCT offered low explanation power, with one reported R² = 15% (H. Li et al., 2016). A model integrating factors from TAM, PCT, and the health belief model showed the strongest explanatory power, with R² = 75% (Cheung et al., 2019).

Postadoption Use (n = 10)

Existing studies explored three types of postadoption use behaviors: (a) continued use (n = 6), which refers to long-term adherence after the initial adoption of wearables; (b) selective use of wearable technology features (n = 3), which indicates exploration and selection of different functions of wearables following adoption; and (c) intermittent discontinuance (n = 1), pointing to an in-between state where people neither continuously used the wearable device nor entirely abandoned it.

Researchers explored the determinants of successful adoption, indicated by continued use (n = 6). Determinants identified were users’ characteristics, such as self-efficacy (Windasari et al., 2021) and health motivations (Deranek et al., 2021; Peng et al., 2021), and technological features such as accuracy, perceived usefulness, and ease of use (Canhoto & Arp, 2017; Lunney et al., 2016; Shin et al., 2017).

Studies also found that users applied selective use (n = 3) postadoption. This line of research generally highlights a need for fit between the motivation and needs of users and the features of wearables to support achieving health-related goals.

Marakhimov and Joo (2017) used coping theory to develop a theoretical model for predicting the extended use of wearables, including exploring and applying more of the available functions of wearables at hand. The model suggested that concerns about privacy and health initiate a coping process, and coping with problems (r = 0.52) and emotions (r = 0.22) significantly predicted extended use (R² = 0.36). Using affordance theory, Jarrahi et al. (2018) identified five groups of users (i.e., aspiring starters, motivation seekers, quantified selfers, curious immobiles, and persistent roamers) and found that the former two types of users interact with both information and motivation aspects of the activity trackers, whereas the latter three only interact with its information affordances. Applying goal content theory, James, Deane et al. (2019) identified three types of exercise goals: intrinsic goals, body-focused extrinsic goals, and social extrinsic goals, and examined how these impact the way users selectively use features of wearable fitness technology. They found that exercisers who had intrinsic goals preferred data features, those who had body-focused goals used data features and social features, and those who had social goals used social features and exercise control features such as goal-setting.

Identifying mechanisms of intermittent discontinuance (n = 1), Shen (2018) found a status in which the adequate expectations on information system use are fulfilled but the desired expectations remain unfulfilled. This is conceptualized as a neutral expectation-disconfirmation based on the expectation-disconfirmation model. It exerts positive effects on neutral satisfaction and attitudinal ambivalence, both of which further have positive effects on intermittent discontinuance.

Health Behavior Change (n = 7)

Studies investigated the psychosocial mechanisms of wearable technologies in promoting health behavior change. Social-cognitive predictors (i.e., self-efficacy, intention, and action planning) partially mediated the relationship between technology interventions, including wearables, and physical activity change (Ratz et al., 2019; Rieder et al., 2021). Health empowerment from self-observation and goal-setting played a mediating role between features of wearables (such as information feedback) and health commitment, which refers to people’s psychological attachment to health goals (Nelson et al., 2016). In addition, perceived companionship fully mediated the relationship between feedback types of wearables (allocentric vs. egocentric, fixed-goal vs. contextual-goal) and health behavior change (Jang & Kim, 2019).

James, Deane et al. (2019) suggested that users’ specific exercise motivations shape the feature sets of wearable fitness technologies used and that the ramifications of effects occurred due to the interaction between motivation types and feature set use. The results suggest that exercisers who are more self-determined (i.e., with intrinsic regulation) have a positive relationship with subjective vitality and that the use of social interaction features and data management features positively moderated this relationship.

Despite the positive role of wearables for promoting health behavior, one study indicated that the use of wearable activity trackers could create a dependency that can harm motivation (Attig & Franke, 2019). Findings suggest that the dependency effect was stronger for participants with high extrinsic motivation for physical activity, a high need for cognitive closure, and low hope of success.

In conclusion, the mediating mechanisms of wearable-based health behavior change are increasing self-efficacy (Ratz et al., 2019; Rieder et al., 2021), empowerment (Nelson et al., 2016), and perceived companionship (Jang & Kim, 2019). Preexisting motivations and individual characteristics have a moderate effect on the relationship between wearable-based intervention and health behavior change (Attig & Franke, 2019; James, Wallace, et al., 2019). In addition, wearable use patterns of following, ignoring, combining, and self-leading bring about different behavioral outcomes (Lehrer et al., 2021).

Individual Outcomes (n = 61)

Studies on individual outcomes looked at behavioral or health outcomes related with individual use of wearables. These were either objectively measured (n = 41) or self-reported (n = 20). Since studies on wearables mostly involved actual use of a wearable device, it made sense that objective measures were more frequently reported.

Studies indicated that the use of wearables is positively associated with self-reported outcomes (n = 20) such as health literacy (Sobko & Brown, 2019), health consciousness (Stiglbauer et al., 2019), and general health status (Ratz et al., 2019). More convincing evidence in the form of objectively measured outcomes (n = 41) was also provided, which indicates that the use of wearable technology could improve physical activity (Pope et al., 2018), sitting posture (Kuo et al., 2019), gait (Ginis et al., 2017), and food consumption (Jasper et al., 2016).

Among these individual-focused outcomes, the amount of objectively measured physical activity was examined most frequently (n = 30), while other outcomes were scarce. Evidence regarding physical activities can be divided into two groups: interventions solely using wearables (n = 14), and multifaceted interventions including various engagement strategies in addition to wearables (n = 16). Multifaceted interventions appeared to be more effective than those that included just the use of wearables, as we found that approximately 80% (n = 13) of the former reported significant improvements in physical activity, whereas only 50% of the latter (n = 7) noted improvements.

In the overall literature on individual outcomes, numerous studies showed mixed results of the effectiveness of wearables for intermediate health outcomes (i.e., health behavior change). However, there is a scarcity of evidence regarding endpoint effectiveness (n = 2). One study showed that the use of wearables could help lower blood pressure, hemoglobin A1c, and low-density lipoprotein cholesterol in patients with uncontrolled hypertension and type 2 diabetes (Frias et al., 2017). Another study showed that a health program based on wearables did not have significant impact on cholesterol or blood pressure (Yu et al., 2017).

Organizational Outcomes (n = 3)

Studies examined wearables for improvements in healthcare organizations from two aspects: healthcare quality (n = 2) and efficiency (n = 1). With regard to healthcare quality (n = 2), one study suggested that using a wearable patient sensor in an intensive care unit significantly helped to reduce the odds of pressure injuries (intervention group = 0.7% vs. control group = 2.3%) through the provision of optimal turning reminders aided by wearable sensors (Pickham et al., 2018). The other demonstrated that the use of wearable cardioverter defibrillators improved clinician adherence with the guideline-directed medical therapy for patients with heart failure (Mirro et al., 2018). For healthcare efficiency (n = 1), one study showed that use of a wearable system (HAIL-CAT [Human Alerting and Interruption Logistics–Clinical Alarm Triage]) for meta-cognitive attention reduced nurses’ response time to important alarms in care delivery from 8.12 to 3.27 minutes (McFarlane et al., 2018).

Leveraging wearable technologies in healthcare organizations showed potential for empowering clinicians to take care of patients with high quality and efficiency. Yet, the evidence of the effectiveness of wearables in the clinical setting is scarce, and further investigations are needed.