Effective and timely care coordination and communication are critical components of efficient and safe patient care [1,2]. Failure in providing coordinated care and communicating patient data is seen as one of the root causes of adverse events such as delays in patient care and deviations from standard medical procedures [3]. The challenges in maintaining effective care coordination and communication are exacerbated when care providers are distributed (eg, located in different places) [4,5]. Over the past 2 decades, many telemedicine systems have been developed to augment remote clinical consults [6-8]. During the COVID-19 pandemic, the need for such systems became more obvious. Most telemedicine systems are implemented on desktops or tablet devices [6,7]. However, these devices have practical limitations: (1) desktop systems have limited portability because they are installed in a fixed location; and (2) tablet device–based systems rely on manual input and control, which can hinder usability [9,10]. These issues could result in limited use of technology in real time, especially during complex care environments and time-critical patient scenarios because they demand the full cognitive attention and physical involvement of care providers [11].

In recent years, the use of smart glasses—a computing device worn as a conventional pair of glasses (Figure 1)—has been gaining momentum in health care because they allow for real-time visual communication in a hands-free manner [12,13]. In particular, smart glasses can present both imagery and textual information within the wearer’s field of view (FOV) through a prism and enable videoconferencing for consults or second opinions via a front-facing camera. Since the introduction of smart glasses to the market, researchers have explored their applicability and usefulness in various medical settings and clinical scenarios [9], such as broadcasting surgeries to facilitate resident teaching [14], recording encounters with patients in wound care [15,16], assessing patients in mass casualty incidents [17], and supporting communication between prehospital and hospital providers [18,19].

As there is a growing interest in using smart glasses to support care coordination and communication across distributed care providers [9,11,20], the aim of this study was to synthesize the knowledge and experiences in this area, understand the benefits and limitations regarding adopting smart glasses as a telemedicine tool, and inform the design of future smart glass applications to better support remote care coordination. We focused on the use of smart glasses in care coordination in various clinical settings (eg, surgical operation, emergency care, and intensive care unit). Our specific research questions were as follows:

These research questions were answered through a systematic literature review covering research within both health care and computer science fields.

Our work contributes the following to the medical informatics community: (1) an in-depth analysis and synthesis of prior research on the use of smart glasses for care coordination and communication; and (2) methodological and design implications for future research on smart glasses to improve distributed care coordination and communication.

Our search started with discussing the search time frame and the most appropriate databases to use as well as search terms with experienced librarians. Using technology keywords such as “smart glasses” and “heads-up display,” along with health care keywords such as “distributed care” and “telemedicine,” a health librarian performed database searches for articles published between January 1, 2000, and March 1, 2022. We chose this time frame to capture the evolvement of this technology (ie, from early concepts such as head-worn displays [21] to smart glasses, which became a well-known concept after the introduction of Google Glass in 2013 [22]). The full list of search terms is presented in Textbox 1. We chose the following databases to cover research within both health care and computer science: ACM Digital Library, Cochrane Library, IEEE Xplore, Ovid MEDLINE, Embase, and Web of Science. A sample search strategy for Ovid MEDLINE is illustrated in Textbox 2. The database searches were set to include only studies published in peer-reviewed journals and conference proceedings in English. Literature reviews, dissertations, posters, and extended abstracts were excluded from the literature search. The retrieved citations were stored and managed using EndNote bibliographic management software (version X9; Clarivate).

We used the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) methodology to search and screen articles [23]. Figure 2 outlines the number of records that were identified, included, and excluded through different phases. More specifically, 5865 articles were identified through database searches, of which 5862 (99.95%) were included for screening after removing duplicates. Article titles were screened first, followed by abstract screening, to identify relevant articles. Of the 5862 articles, after screening of article titles, we excluded 5341 (91.11%); of the remaining 521 studies, 446 (85.6%), were excluded, leaving 75 (14.4%) for full-text review. After reviewing the full text of these 75 articles, we deemed 21 (28%) to be eligible for this systematic review.

Three authors (EB, KJ, and PG) independently screened all papers through the paper stack and selected relevant papers for inclusion. Two senior researchers (ZZ and MO) oversaw the whole article review and selection process. Any conflicts in selection decisions were resolved through discussion among all the authors during weekly group research meetings. The inclusion criteria were peer-reviewed articles that reported the use or testing of any smart glass technology and accompanying software in the context of communication and collaboration across distributed care providers. Articles were excluded if they only reported the use of smart glasses by an individual or in a collocated clinical setting or if they did not provide adequate supporting information, such as what clinical setting the smart glasses were used in and who used the technology.

Guided by the research questions of this study, 2 authors (KJ and EB) used a Microsoft Excel spreadsheet to extract, collate, and summarize data from the included studies, such as the country where the study was conducted, study objectives and scope, clinical scenarios, system evaluation methods, technology specifics, barriers and challenges, and a summary of study findings. Textbox 3 summarizes these data fields and their brief definitions. In addition to extracting the aforementioned metadata, we also assessed the technology readiness levels (TRLs) [24] of the systems tested in the reviewed studies. There are 9 different TRLs, ranging from level 1 (scientific knowledge generated underpinning hardware and software technology) to level 9 (actual system “flight proven” through successful mission operations). Two authors (KJ and EB) followed the metrics proposed in the study by Engel et al [25] and independently assessed TRLs for each system. They then compared and discussed their TRL evaluations until they reached agreement.

Two senior researchers (ZZ and MO) reviewed all the articles and analyses as a verification step. The research team met regularly to discuss the results. We performed the data analysis iteratively (ie, we went back and forth as more knowledge was obtained), as suggested by prior work [11,26]. A meta-analysis of the study results was not considered in this work owing to the heterogeneity of the study designs and results.

In the following section, we report information that was synthesized from the reviewed articles, including characteristics of the selected studies, system architecture and features, TRLs of the reviewed systems, system evaluation methods, and care providers’ perceived benefits and challenges of using and adopting smart glasses for distributed care coordination.

Of the 21 reviewed articles, 10 (48%) were conducted in the United States [18,19,27-34], and 2 (10%) were conducted for surgical teleproctoring between high-income countries and low- and middle-income countries (LMICs), such as between surgeons in the United States and Mozambique [35] and between experienced surgeons recruited from the United States and Germany and novice surgeons in Brazil and Paraguay [36]. The remaining studies (9/21, 43%) were conducted in different countries, such as Spain [37], China [38], Germany [39], France [40], Italy [41], Switzerland [42], Malaysia [43], South Korea [44], and Republic of the Congo [45]. The reviewed studies were conducted to assess the feasibility, effectiveness, and user experience of smart glasses in supporting remote patient evaluation and care procedure operation in a particular medical domain. The study objectives, along with major findings for each reviewed article, are presented in Multimedia Appendix 1 [18,19,27-45].

The clinical foci in these 21 papers vary: 9 (43%) focused on surgical settings [29,30,33-38,44], whereas 6 (29%) focused on the prehospital or emergency medical services domain [18,19,28,31,39,42]. The remaining studies (6/21, 29%) focused on intensive care [40,43], toxicology [27], ophthalmology [32], pediatric cardiology [41], and general medicine [45].

The scope and purpose of the use of smart glasses among these studies vary. As shown in Figure 3A, the majority of the reviewed studies (16/21, 76%) used smart glasses to enable remote patient care and evaluation [18,19,27,28,30-32,37-45]. Of these 16 studies, 8 (50%) [27,28,30,32,37,38,43,45] tested smart glasses with real patients, 6 (38%) [18,19,31,39,40,44] conducted system testing in a simulated environment, and 2 (13%) [41,42] did not specify how the device was tested. The remaining studies (5/21, 24%) [29,33-36] leveraged smart glasses for training and teleproctoring purposes; of these 5 studies, 4 (80%) [29,34-36] tested the device with real patients, whereas 1 (20%) [33] tested the device in a simulated environment.

The reviewed articles were published between 2014 and 2021 (Figure 3B). It is noticeable that almost half of the reviewed articles (9/21, 43%) were published within the first 3 years of the release of Google Glass [22]. Subsequently, the number of studies on the use of smart glasses for supporting distributed care decreased until 2021. One possible explanation for this finding is that the use of smart glasses regained momentum right after the outbreak of the COVID-19 pandemic as researchers started exploring smart glass use to enable medical personnel to participate in remote assessment and consultation, with the aim of safeguarding patients and health care providers during the pandemic.

Although the system architecture implemented in each study varied, there were some similarities across the reviewed studies. Typically, there are two types of technology setups on the local site: (1) smart glasses are connected to a Wi-Fi network, a Wi-Fi hotspot, or a mobile router to directly stream the first-person point-of-view to a remote consultant (Figure 4A); or (2) smart glasses are connected to a smartphone or a laptop via Bluetooth or Wi-Fi for video streaming and audio transmission (Figure 4B). The first approach was adopted by 52% (11/21) of the studies [19,27,28,31,32,36,38,39,41,43,44], and the second approach was used in 33% (7/21) of the studies [18,29,35,37,40,42,45]; for example, in the study by Diaka et al [45], the smart glasses were designed as an extension of a smartphone, which meant that the local wearer needed to initiate the call on the smartphone. Regardless of the system implementation method on the local site, the remote experts were usually equipped with either a computer or a mobile device (eg, a tablet device) to review and access the video stream and other multimedia data shared by the local medical practitioner (Figure 4). However, it is worth mentioning that in the study by Brewer et al [33], where smart glasses were used for surgical training, the remote expert (trainer) also wore a pair of smart glasses to view the video streamed from the learner.

As shown in Figure 5A, the reported brands of smart glasses in these studies included Google Glass [18,19,27-30,33-36,40,42,44], Vuzix [38,43], Iristick [37,45], Pivothead Original Series [32], Intel Recon Jet [31], and Epson Moverio BT-200 [41]. Google Glass was the most frequently used smart glass device (13/20, 65%). Another interesting observation is that all of the studies (21/21, 100%) used off-the-shelf, commercialized videoconferencing software (Figure 5B) such as Pristine Eyesight [19,27], AMA XpertEye [28,35], Livestream [18,36], WebRTC (enabled by Google) [42,44], Livecast Media [38], Skype [29], CrowdOptic [33], Google Hangout [34], and Polycom RealPresence Group 500 [32]. Most of the videoconferencing software used was compliant with the Health Insurance Portability and Accountability Act (HIPAA) rules, except in the case of the study by Cicero et al [18], where the researchers only tested the use of smart glasses in a simulated environment (real patient care was not involved).

Although there was variation in the application scopes and domains, there were some common software features across the reviewed studies (Textbox 4). Real-time synchronous video and audio streaming from the local smart glass wearer to the remote consultant is the most common feature among the studies (19/21, 90%). In the case of the exceptions (2/21, 10%), because of technical limitations (eg, limited internet connection), the study by Gupta et al [30] first recorded patient care and evaluation using smart glasses and then transmitted the recordings to remote experts at a later time to simulate real-time telemedicine consults, whereas in the study by Hashimoto et al [34], researchers used Google Glass and an Apple iPhone to capture videos of a surgical operation and compared the video quality and its adequacy for safe use in telementoring.

Another noteworthy feature is enabling imagery and text-based remote guidance and annotation; for example, the remote consultant can annotate images captured from the live stream and project them back onto the local glass wearer’s visual field [35,37]. In 19% (4/21) of the studies [19,27,36,44], the remote consultant could use the texting feature to type messages that could be projected onto the smart glass display. These annotation features provide the remote consultant with more channels (in addition to audio and video) to direct and guide local medical practitioners to perform critical procedures.

Augmented reality (AR)—a technique that can enhance an individual’s visual experience of the real world through the integration of digital visual elements—was also tested in several studies. In Ponce et al [29], for example, AR enabled a remote surgeon to insert their hands or instruments virtually into the visual field of the local surgeon who wore smart glasses for real-time guidance, training, and assistance as needed. In another study [41], a remote specialist used AR-based markers to guide the execution of an echocardiographic examination performed by a local operator. The markers were overlaid on the ultrasound device and could be seen through the screen of the local operator’s smart glasses.

Other features of smart glasses reported in the studies included zooming in and out of the live stream video [35]; using voice commands [27,28,30,31] or head movements [27] to control, and interact with, the smart glass device; taking photographs [19,30,31,35]; automatically detecting the geographic location of on-site medical teams with the built-in GPS [31]; and presenting prehospital triage algorithm on the glass screen for decision support during mass casualty incidents [39].

On the basis of our analysis, we found that the TRLs of all the systems used or tested in the reviewed studies ranged between 7 and 9. Our TRL assessment for each system is visualized in Figure 6 [18,19,27-45]. The reasoning for our assessment is summarized in Multimedia Appendix 2 [18,19,27-45].

The systems in 24% (5/21) of the studies [18,34,40-42] have a TRL of 7, which indicates that the technology is in the form of a high-fidelity prototype and has all key functionality available for demonstration and test; for example, in the study by Widmer and Müller [42], the Google Glass device on the local site was set up to connect with a computer application on the remote site for teleconsultation. This integrated system was only preliminarily tested by the research team but not in a simulated or real environment (a criterion for TRL 8); thus, its TRL was set to 7. It is worth mentioning that of these 5 studies, 3 (60%) [18,40,41] tested smart glasses in simulated scenarios; however, there were several reasons for their failure to meet the criteria for TRL 8, such as using non–HIPAA-compliant videoconferencing software, testing the technology with only 1 volunteer, or not fully integrating smart glasses with the network and remote devices.

The majority of the studies (15/21, 71%) [19,27-33,35-39,43,44] tested or used systems that met the criteria for TRL 8, indicating that they are actual systems in their final configuration and have been fully developed and tested in either simulated or real operational scenarios. However, these studies provided limited information regarding some criteria for TRL 9, such as whether the system had been fully integrated with other operational hardware and software systems (eg, database and hospital IT infrastructure), whether all system documentation had been completed, whether training on system use was available, and whether engineering support team was in place. Without such information, it is difficult to assess the readiness of these systems for large-scale deployment.

In comparison, only the system in the study by Diaka et al [45] was assessed to have a TRL of 9 because the system had been successfully operated on actual missions and tasks in the operational environment for a relatively long time (ie, more than a year). Furthermore, the system was fully integrated with other operational software, hardware, and network devices, as well as care delivery services (eg, moto-ambulances to facilitate patient referrals after teleconsultation).

In this work, we conducted a systematic review of studies focused on the use and application of smart glasses in supporting care coordination and communication among distributed medical teams. Of the 5862 papers included for screening, only 21 (0.36%) met our criteria, highlighting the paucity of studies examining the feasibility, effectiveness, and user experience of using smart glasses as a telemedicine tool. Furthermore, the studies were mostly conducted in the United States and a few other high-income countries (eg, Italy, Germany, and France). One possible explanation is that smart glass technology is costly, hindering its adoption in LMICs and low-resource settings. However, 14% (3/21) of the reviewed studies [35,36,45] revealed the substantial benefits that smart glasses could bring to LMICs and rural areas, such as providing remote training and mentoring and more accurate instructions to the field medical practitioners in low-resource settings who otherwise have limited access to remote experts. Given such benefits, more future work is needed to expand the research of smart glasses to LMICs.

Another interesting observation is that all the reviewed studies (21/21, 100%) only used off-the-shelf hardware and software without involving users in the system design process. Prior work has suggested that it is critical to involve users and understand user requirements in the early phase of system development to identify and address potential usability and technical issues [6,46,47]. In addition, regarding the methodology for eliciting user opinions, out of 15 studies conducted user evaluation, 33% (5/15) of them did not specify what questions they asked, how the questionnaire was developed, and what procedure was followed. Despite the user-friendliness of health care information technology being a determinant factor for user adoption and acceptance [48,49], the usability of smart glasses was neglected by most of the studies (19/21, 90%), with only the studies by Broach et al [19] and Demir et al [31] specifically examining this aspect. These facts highlight the need to adopt a user-centered design approach in the development of smart glass technology by placing users at the center of the system design process from inception to implementation and deployment.

A similar concern is that a few of the reviewed studies (4/21, 19%) only recruited a small number of study participants (eg, 2 health care professionals) to participate in their user studies (eg, survey or interview). In addition, some of the studies (5/21, 24%) did not report the details of their user research, including the number of participants. These findings may suggest that the important role of user research was not recognized in some of the reviewed studies (9/21, 43%), and their results might not be generalizable because of the limited number of study participants. Given these study limitations, we argue that involving human-computer interaction researchers in such type of research and establishing close collaborations between these researchers and health care domain experts are critical and much needed, as demonstrated in the study by Schlosser et al [50].

Finally, almost all of the reviewed studies (20/21, 95%) focused on evaluating the smart glass technology either from a technical perspective or a clinical perspective, while neglecting other important factors that could substantially affect the use and adoption of this technology, such as workflow, teamwork, policies, and organizational cultures. As prior work has argued [51], an ongoing challenge to the successful implementation and deployment of health IT (HIT) interventions is to operationalize their use within the workflow of a complex health care system; for example, a new technology could disrupt current clinical work, causing not only frustrations for medical providers but also patient safety issues [52-54]. When this problem occurs, not surprisingly, medical practitioners are left with no choice but to bypass the technology or adopt informal, low-tech, potentially unsafe workarounds that deviate from the formal protocol [55,56]. As such, researchers have highlighted the importance of examining the design, use, and application of HIT interventions through the lens of a sociotechnical perspective [55-57]. This approach allows researchers and practitioners to understand the complex interrelations between various social and technical elements of systems that are equally important in determining the success of HIT adoption in a health care organization. In line with this argument, we believe that more research adopting a sociotechnical model [51,58] is needed to investigate the factors (eg, human-computer interaction, workflow and communication, internal organizational features, and external rules) that contribute to the uptake of smart glasses in routine use.

The reviewed studies revealed a set of challenges and barriers to adopting and using smart glasses in practice; for example, a commonly cited technical challenge is internet connection quality—smart glasses rely on a high-bandwidth internet network for streaming videos and transmitting other visual media data (eg, high-resolution pictures, texts, and augmented objects). However, this technical requirement could be challenging to fulfill, especially in low-resource or out-of-hospital settings [59]. With the rapid development of 5G technology, this technical barrier might be overcome in the near future; for example, a study [60] showed that 5G technology could not only enable safe and efficient complex surgical procedures during telementored surgery but also lead to a very high degree of surgical team satisfaction. In addition to internet connections, other technical improvements suggested by the reviewed studies include increasing the memory space of smart glasses to store more information, adding autofocus and stabilization features to the smart glass camera, and improving the camera resolution [35].

Human factors and usability issues make up another set of important considerations for smart glass designers and developers; for example, the difference in line of sight between the local medical practitioner and remote consultant impeded the remote consultant from seeing exactly what the smart glass wearer’s eyes were fixed on. In addition, the limited FOV further complicated the video transmission to the remote experts. One reviewed study [44] experimented by attaching a mirror to the smart glass to increase the FOV of the local practitioner by transmitting both the wearer’s front view and their hand operations below the camera to the remote experts. However, the video received on the other end by the experts was deemed confusing. Another viable solution suggested by prior work [59] is using more advanced mounting techniques to make sure that the smart glass can sit steadily on the wearer’s head to align their visual field with the camera range. Another interesting issue brought out by a few of the studies (5/21, 24%) was the necessity of enhancing user interactions with the smart glass, such as offering more hands-free interaction mechanisms (eg, using head movements to control the device) [35] and enabling the user to zoom in and out during video streaming as well as pan the image [38].

Current smart glass applications are stand-alone and limit their potential. The data collected and transferred through smart glasses can best benefit patient care tasks if they can be incorporated into, and fully integrated with, other HITs such as electronic health records or clinical decision support systems. Interoperability issues (eg, standardized terminology) should be considered when deploying and integrating smart glasses into complex health care systems.

Other important design considerations that need full attention for developing and deploying the smart glass technology include (1) ensuring that the software is compliant with HIPAA requirements to protect patient privacy and data security, (2) integrating smart glasses into the workflow to minimize the disruption to medical practitioners’ work, and (3) providing sufficient training to end users.

Defining the search keywords was difficult. To generate a comprehensive and relevant list of keywords, we iteratively discussed and selected the keywords for the search based on suggestions from the health librarian and a review of systematic review articles regarding smart glasses. Another limitation is that we did not assess the quality or impact of the results from the included articles. A meta-analysis was not feasible because of the heterogeneity of the study designs and results.

Smart glasses were found to be an acceptable and feasible tool in enabling visual communication and information sharing among distributed medical teams. Despite the high potential of this novel technology, the reviewed articles pointed out a set of challenges that need to be addressed before the wide deployment of this technology in complex health care systems. Thoughtful system design involving end users from the beginning and improved hardware and software reliability are needed to improve the usefulness and usability of smart glasses for medical practitioners [11,59]. We suggest that more user-centered design and evaluation research is needed to examine and evaluate medical professionals’ needs and perceptions and determine how to design smart glass technology to meet their needs. In addition, more research is required to elucidate how smart glasses affect the workflow of medical professionals in complex care environments.