With the emergence of various wearable devices in recent years, the concept and statement of “smart medical care” are gradually emerging in medical innovation. The development of smart medical care has a long history. In addition, with the advancement and rapid rise of the internet of things (IoT) technology, a large amount of medical information has been exchanged and analyzed, which has become the basis of medical big data. Artificial intelligence, which has developed rapidly in recent years, has been introduced as an inductive use of these data. After the combination of the IoT and artificial intelligence, instant mobile medical care emerged, which is the core concept of smart medical care.

Wearable devices can detect the physical condition, use real-time perception, and compare and analyze a large amount of data for analysis, interpretation, and response and can then select the most appropriate current processing and support. Through smart medical care, many dilemmas faced by the current medical system have been resolved. The global market for wearable medical devices is expected to increase from US $6.22 billion in 2017 to nearly US$ 14.41 billion in 2022 at a compound annual growth rate of 18.3% (2017-2022) [1]. Furthermore, the emerging market demand introduced by smart health care is also a big business opportunity.

Further, according to Gartner’s latest forecast [2], by 2020, global end user spending on wearable devices will reach US $51.545 billion, up 27% from US $40.581 billion in 2019. Among them, consumers will spend the most on smartwatches and smart clothing, growing 34% and 52%, respectively. In the past few years, the improvement of sensor accuracy, the development of miniaturization, and better user data protection have made more consumers willing to buy wearable devices. As for hardware manufacturers, they are focusing on sensors that are smaller and smarter, so that the sensors built into wearable devices can obtain more accurate readings, and more usage examples continue to appear. Previous literature has focused on a single form of smart medical service, such as discussing the application of wearable medical services from the perspective of developers [3]. There are few studies considering wearable medical devices from the perspective of users. To fill the abovementioned research gaps, this study developed and validated empirically a model that explicates users’ intentions to use wearable medical devices. Specifically, it revisits a popular contemporary adoption theory (unified theory of acceptance and use of technology [UTAUT] [4]) by augmenting it to better capture wearable medical device environments. Recently, Zhou [5] added a health consciousness construct to the UTAUT in a wearable medical device context. Zahir and Gharleghi [6] also effectively introduced an innovation-related construct (trust), which influences users to adopt the technology. Thus, these two constructs are especially relevant when identifying users’ characteristics regarding the adoption of information technology (IT).

Hence, this study augments the application of the UTAUT and adds two individual factors (health consciousness and trust) to explain users’ intentions regarding wearable medical devices. The purpose of this study was to combine the UTAUT and the two specific factors to improve the IT adoption model and explain the users’ intentions for wearable medical devices.

The conceptual model for the study was developed from the researcher’s view of the interactions that could exist between the variables of the study based on a review of the literature. The model proposes a direct relationship between the independent variables and the dependent variable. Specifically, it is assumed that there is a relationship between PE and the use of wearable medical devices. In addition, there could be a link between EE and the use of wearable medical devices. It is also evident from the model that a relationship could be proposed between FC and the use of wearable medical devices. In addition, the model also seeks to test the influence of the three independent variables on the dependent variable (Figure 1).

For data collection, this study developed a self-administered online survey. Measurement scales for all construct items were taken from existing scales based on prior work [4,20], with modified wordings to adapt the items to the topic area. In the pretest phase, the questionnaire was reviewed by a small group of IS faculty and management students. The scales were modified as a result of their suggestions. The questionnaire was then tested with a sample of medical and business school students and personnel. This resulted in further modifications to the questions. The purposes of these pretests were to confirm that relevant aspects were included and to enhance the clarity and readability of the questionnaire.

To ensure the content validity of the scales used, the items selected should represent the concept around which generalizations are to be made. Items selected for the constructs were therefore largely adapted from prior studies [4] to ensure content validity. In this study, the constructs of the UTAUT were taken from the study by Venkatesh et al [4] and modified to reflect the utility of wearable medical devices, whereas the constructs of health consciousness and trust were taken from the studies by Ahadzadeh et al [15] and Safa and Solms [21].

The participants were instructed to rate each item of the dependent variables on a 5-point Likert scale from 1 (strongly disagree) to 5 (strongly agree).

This study posted an electronic survey through an online survey platform (Survey Cake) to obtain a sample from the mass population of wearable medical device users. The wearable medical devices included devices and technologies (eg, wearable glucose monitoring and drug delivery devices, activity monitors, smart clothing, smart equipment, wearable vital sign monitors, and smartwatches). Using Facebook, a popular social networking site, potential participants were chosen to complete surveys. After clicking on the link and entering the questionnaire website, participants were considered wearable medical device users or potential users.

An online self-administered survey questionnaire was considered an appropriate instrument to identify wearable medical device users. All questions in the questionnaire were measured on a 5-point Likert scale, ranging from “1” (strongly disagree) to “5” (strongly agree). The time required to complete the questionnaire was almost 3 to 5 minutes. The final questionnaire of items is presented in Multimedia Appendix 1 [4,20,21].

All subjects participated in the study voluntarily during the period from July 20 to August 20, 2019. There were 452 participants overall (252 male and 200 female participants). The mean age was 47.8 years, and participants aged 39 to 55 years accounted for 53.1% (240/452) of the study sample. Most of the participants (344/452, 76.1%) stated that they were familiar with the term “wearable medical devices” prior to completing the survey.

Structural equations among latent constructs were examined to test the conceptual structural equation model (SEM). The SEM was used to analyze causal models and simultaneously estimate a series of interrelated dependence relationships. Thus, data analysis was carried out using structural estimation modeling. Before this study tested the research model, SPSS 25.0 for Windows (IBM Corp) was used to show the important descriptive information on demographic variables, including participant characteristics such as gender, age, and educational background. This information also included behaviors related to the use of wearable medical devices, such as the time spent on the internet, the preferred online medical platform provider, and the frequency of using wearable medical devices. Model evaluation involved a two-step analysis [22] using the software IBM Amos 21.0. For this purpose, the author first built a measurement model using confirmatory factor analysis for the model to check its fit and then built the SEM and examined the hypothesized causal paths among the constructs by performing a simultaneous test. This helped to observe whether the conceptual framework had provided an acceptable fit to the empirical data.

The validity of the measurement model was evaluated by investigating convergent validity, discriminant validity, and reliability. Structural equation modeling has been used to evaluate the plan’s research model and hypotheses. Simultaneously, for assessing the reliability of measurement items, this research computed composite construct reliability coefficients. Therefore, all the average variances extracted exceeded 0.50, all composite reliabilities were larger than 0.70, the factor loadings of all items exceeded the recommended level of 0.60, and all values were significant at .001, demonstrating that the scales had good convergent validity. In addition, the Cronbach α of the seven constructs ranged from .81 to .89. All composite reliabilities were larger than 0.70, displaying good reliability [23]. The results confirmed good reliability (Table 1).

Discriminant validity is shown when (1) measurement items load more strongly on their assigned construct than on the other constructs in a confirmatory factor analysis and (2) the square root of the average variance extracted of a construct is larger than its correlations with the other constructs [24]. To test the discriminant validity, this research computed the square root of the average variance extracted and factor correlation coefficients. For each factor, the square root of the average variance extracted should be greater than its correlation coefficients with other factors to show that the scale has a worthy discriminant validity [25]. As shown in Table 2, all constructs had an average variance extracted value higher than the threshold of 0.50, confirming the convergent validity of the constructs.

After the measurement model was satisfied, the structural model was evaluated, and it was well converged. The results investigated the chi-square of the structural model, ratio of chi-square to df, goodness-of-fit index, adjusted goodness-of-fit index, normed fit index, comparative fit index, root mean square residual, and root mean square error of approximation. Table 3 presents the model fit indicators with their respective criteria as follows: (1) the comparative fit index was 0.91 (greater than 0.90), (2) the root mean squared error of approximation was 0.03 (smaller than 0.08), and (3) the goodness-of-fit index was 0.93 (greater than 0.90) [25-29]. These indicators were acceptable and showed good fit of the model to the data.

Significance was determined by running bootstrapping calculations with 352 samples and no sign variation. Four paths were relevant as shown in Figure 2.

Figure 2 shows the graphic description and the numerical results of the path coefficients. There were significant effects by PE (β=.42; P<.001), EE (β=.34; P<.001), SI (β=.46; P<.001), FC (β=.23; P<.001), health consciousness (β=.68; P<.001), and trust (β=.48; P<.001). The coefficients of these variables were statistically significant (P<.001) and had the expected signs (Figure 2). All hypotheses were supported.

The results of this study provide several implications for researchers and practitioners. First, the results reveal the magnitudes of the impacts of the variables in the well-accepted revised UTAUT model in the context of the medical industry, particularly in the setting of wearable medical devices. Indeed, PE, EE, SI, and FC lead to positive intentions to use wearable medical devices, supporting H1, H2, H3, and H4.

Second, the impact of SI on adoption intention was more than that of FC, PE, EE, health consciousness, and trust, which is highly relevant in explaining the use of wearable medical devices. This implies that SI is an important factor affecting technology usage intention, that is, SI has positive effects on the intention of using wearable mobile devices. This finding differs from that of most studies on new health care technology acceptance and could reflect the culture, regulations, or rules in the Chinese social context. The result is also consistent with the findings of Ye et al [30]. Compared with another report by the author [31], PE has more positive effects on the intention to use library apps than UTAUT factors. Perhaps in different research backgrounds, the explanatory power of each factor of UTAUT would also be different.

Finally, this study modified the UTAUT by including constructs from health consciousness and trust. H5 and H6 were supported. The path coefficients were relevant, so the additional effects of health consciousness and trust were present. In particular, the degrees of health consciousness and trust were positive for strongly influencing the effects of the usage intention of wearable medical devices. In other words, this study demonstrated that higher health consciousness and trust can lead to much stronger intentions for using wearable medical devices. This study introduced health consciousness and trust as predictors in the Chinese social context to reflect the health care context, and this result is consisted with the findings in the studies by Dou et al [32] and Andrews et al [33].

The implications of this study for practice are twofold. One practical implication is that based on the findings of the research model, service providers can make an effort to design a frequently well-used interface in order to enhance users’ PE, EE, and FC regarding the intentions of using wearable medical devices. In addition to the roles of PE, EE, and FC in usage intentions, SI has a positive effect on the intentions of using wearable medical devices. Service providers may still encourage users to spread positive word-of-mouth information (eg, positive ratings) to increase peer use. Another practical implication is that the research presented in this paper demonstrates the effects of health consciousness and trust in the use of wearable medical devices. Service providers can consider how to develop a very complicated device that takes into account an individual’ s ability and cognition in order to better match the wearable medical device user’s needs.

Although the research findings contribute to the practice of marketing, the study is characterized by several limitations that may provide opportunities for future research. One limitation of this study is that as the sample was obtained by considering wearable devices or websites, the number of participants aged above 50 years was relatively low. Their behavior might differ somewhat from the population average, and this may have biased the results. Another limitation is that our study was limited to the customer base of one country. Further research is needed to examine differences in the effects of consumer characteristics across cultures.