Serious mental illnesses (SMIs) impact nearly 15.4 million adults [1], are challenging to treat, and require years of monitoring and adjustments in treatment. Diagnoses of SMI have been linked to additional life challenges, including job loss [2], homelessness [3], incarceration [4], hospitalization [5,6], or suicide [7]. Clinician visits are infrequent, and illness exacerbations often occur with no clinician awareness in real time, leaving no opportunity to adjust treatment. Methods are needed for monitoring mental health and behavioral status of individuals with SMI to support outreach, care coordination, and treatment. Behavioral phenotypes have been developed from mobile data, and wearable devices, which are now common in the population and are well-suited to produce these data and support clinical services. However, it is not yet clear whether people with SMI will accept and are interested in engaging with wearable devices and mobile monitoring to support care delivery.

Individuals with SMI experience several health comorbidities [8], like cardiovascular disease [9], and have higher mortality rates than the general population [10], with some estimating about 2 to 3 times higher [11]. In addition, SMI has been associated with high financial burdens on society through increased emergency service use [12] and has been linked to high rates of disability [13]. Medication is a common treatment for SMI symptoms; however, complex medication regimens present challenges to adherence for patients, and many antipsychotic medications are associated with adverse side effects that may exacerbate or create additional health challenges [14]. Furthermore, this population also experiences additional and systemic barriers, such as stigma, poor coordination of treatment, and lack of interdisciplinary care for physical and mental health symptoms, which can contribute to mistrust in health care systems [15]. Exploring the ways mobile-sensing devices can be used to benefit treatment outcomes may address multiple overlapping barriers, risks, and rapid, unpredictable changes in symptoms.

Past research has shown both increased use of and favorable attitudes toward mobile devices, particularly cell phones, among individuals with SMI [16,17], including Veterans with SMI [18], making the possibility of integrating mobile-monitoring devices and their partner cell phone apps into current health care treatments feasible for this population. Incorporating mobile health interventions via smartphones can improve treatment outcomes [19] and intervention implementation through accessible web-based therapies, leading to improvements in SMI symptoms, reductions in hospitalizations, and increases in social connectedness [20]. In addition, the use of cell phone apps has been found to increase patient knowledge of their health conditions [21], and this additional information may be particularly helpful in allowing individuals with SMI to advocate for their needs and monitor symptoms, as well as give providers more insight into their overall health, which could improve treatment planning and care coordination.

Research conducted among Veterans with SMI shows that remote sensing devices can be used to effectively monitor sleep and physical activity data [22]; however, while this research supports the efficacy of using these devices partnered with intensive support, further study is needed to examine their effectiveness when incorporated into usual care. Changes in health behaviors have been found to be correlated with the onset of psychiatric symptoms, such as abnormal sleep patterns [23] and changes in physical activity levels [24] before experiencing a relapse or a re-emergence of positive symptoms. Therefore, by monitoring these and similar health behaviors through wearable mobile-monitoring devices, patient-generated health data (PGHD) could potentially be used to predict relapses in individuals with SMI. Furthermore, if patients were able to self-monitor their health behaviors with wearable devices, they may be able to seek assistance before a relapse occurs. Additionally, if the health information from wearable devices is shared with patients’ providers, providers can track behavior changes between medical visits, and proactively reach out to patients, and, if needed, adjust treatments to prevent a lapse in symptoms. Research in this area found that clinicians are open to implementing the use of mobile-sensing technology into health care treatment [25,26]. Similarly, individuals with SMI are open to the idea of self-monitoring technology [27]. Further study is needed to better assess potential concerns of mobile-monitoring devices among this population surrounding privacy, location tracking, sharing PGHD with clinicians, as well as discovering ways to improve the devices for individuals with SMI.

In this study, we used qualitative methods to examine the acceptability and feasibility of using mobile-sensing technology to inform and improve health care outcomes for individuals with SMI. We were interested in understanding whether participants found PGHD useful and potential barriers to data use and integration in their health care treatment. This study expands our knowledge of the thoughts and perceptions of individuals with SMI toward mobile-sensing technology and incorporating wearables into their daily lives, which can inform future research exploring the integration of PGHD collected from these devices in a clinical setting to improve treatment outcomes.

This study served as a qualitative arm of a larger pilot study [28] exploring how passive monitoring through a mobile-sensing cell phone app can be used to improve health care treatment and outcomes for individuals with SMI. To be eligible to participate, Veterans needed to receive care at the Greater Los Angeles Department of Veterans Affairs (GLA VA) and have received a SMI diagnosis of either schizophrenia, schizoaffective disorder, or bipolar disorder, or have received a diagnosis of posttraumatic stress disorder (PTSD) and be on a qualifying antipsychotic medication. Research staff screened patients’ records using the Department of Veterans Affairs (VA’s) electronic health record system, and eligible participants were recruited via letter and follow-up phone calls. For the larger study, a total of 96 participants (n=96) were enrolled (535 Veterans were approached, 197 were not interested, 229 were unable to contact, 13 were not eligible due to an early version of the project’s app being incompatible with Android smartphones).

Individuals who participated in qualitative interviews were randomly selected out of the total 96 Veterans in the larger arm of the project to participate in the qualitative portion of the study using wearable mobile-monitoring devices. A total of 15 Veterans (n=15) were approached, and all consented to be interviewed. These participants were provided a wearable health and fitness tracking device (a Fitbit watch, either an Inspire 2 or Inspire 3 version). Fitbit devices were deemed appropriate for this project as they assess the study’s domains of interest (eg, stress, sleep, and physical activity), are compatible with both Android and iPhone cellular devices, are relatively affordable, and are market-leaders for mobile sensing devices. At the time of the interviews, all participants had been using their wearable device for a period ranging from 2 weeks to 1 month.

All Fitbit devices come with an accompanying app that displays the user’s longitudinal health data collected from the wearable device. The app also provides access to mindfulness exercises, physical fitness routines, and allows users to input additional health information, such as food and water intake. Upon providing individuals with their watch, research staff downloaded the Fitbit app to participants’ cell phones and briefly demonstrated how the app can be used along with the wearable device itself.

Semistructured interviews were conducted by research staff. Three interviewers (AF, LH, and RC) from the study team interviewed participants individually using a semistructured interview guide. Two interviewers (AF and LH) self-identified as female and one as male (RC). Two interviewers (LH and RC) were employed at the GLA VA as qualitative analysts and held master’s-level degrees in experimental psychology and social work, and one (AF) was employed as a research coordinator and held a bachelor’s degree in psychology. All interviewers had training in qualitative methods, previous experience conducting interviews, and prior involvement working with Veterans with SMI at GLA VA.

The development of the interview guide was informed by the Technology Acceptance Model [29] and focused on perceptions of individuals with SMI toward wearable mobile-sensing devices. Interviews focused on the acceptability, feasibility, perceived usefulness, and ease of use of wearable mobile-monitoring devices for this population. In addition, interviews probed how these devices and accompanying phone app could be improved to enhance usability and overall experiences for participants. The interview guide was pilot tested by research staff before data collection and then adjusted to fit the desired length of the interviews.

Interviews lasted between 15 and 30 minutes and were conducted either in person or over the phone in private office spaces with only the interviewer and participant present. All interviews were audio-recorded and then professionally transcribed. Interviews were conducted until saturation was achieved. Saturation was defined as the point at which no new themes emerged from 3 consecutive interviews. To assess saturation, transcripts were reviewed iteratively by the research team to track the emergence of new themes and codes. After three successive interviews produced no additional themes, data collection was discontinued.

Rapid qualitative analysis [30] was used to identify prominent findings established to be informative of the perceptions of individuals with SMI toward wearable mobile-sensing devices. A rapid qualitative analysis approach to reviewing interview transcripts was deemed appropriate as the primary goal was to gain overall feedback from Veterans, as well as elicit insights to targeted questions designed to broaden our understanding of the experiences of participants with SMI when using these devices. Four members of the research team (AF, JC, LH, and RC) developed a draft codebook based on thematic analysis, including a priori codes derived from the qualitative interview guide. Through iterative review of transcripts, the research team identified and incorporated emergent themes, refining the codebook to reflect both predefined and data-driven themes [31]. Two team members (AF and JC) then independently coded the same 2 transcripts using Atlas.ti 24 software (Lumivero) and met afterward to compare and discuss the results and modify the codebook based on their findings. Any coding discrepancies were reconciled by consensus. These two authors (AF and JC) then separately coded identical transcripts once more using the refined codebook and met again to compare coding and establish strong intercoder reliability. No discrepancies in coding were found. All remaining interviews were then coded by AF and JC. Findings were organized using emergent themes from interviews.

This research was approved by the Veterans Health Administration’s Institutional Review Board (2018‐020118). Written informed consent was obtained from participants upon enrollment into the parent study, and verbal consent was also obtained from participants before the start of each qualitative interview. This study used deidentified data to safeguard participant information. Participants received US $10 in compensation for completing qualitative interviews. Reporting of qualitative research in this manuscript is aligned with guidelines for qualitative research in informatics [32].

Ages of participants ranged from 34 to 69 years with a mean age of 51.7 (SD 11.9) years. Four of the participants self-identified as women and 11 participants self-identified as men. Five participants were diagnosed with a schizophrenia spectrum disorder, 6 were diagnosed with bipolar disorder, and 4 were diagnosed with PTSD. In addition, 8 participants held diagnoses for 2 qualifying SMIs, 6 with a combination of bipolar disorder and PTSD, and 2 with a combination of a schizophrenia spectrum disorder and PTSD (Table 1).

Participants with SMI in this sample generally held positive perceptions toward wearable mobile-monitoring devices and were able to incorporate wearing the health and fitness trackers into their daily lives, many noting that the passive data collection was convenient. Qualitative data showed that participants used the devices to track various aspects of their health, such as sleep, heart rate, stress, and physical activity, and found that devices gave them a greater understanding of their overall health. Some participants experienced technical difficulties when using the wearables and their partner phone app, which may have limited their use of the devices. Additionally, participants suggested that some physical changes to the wearable devices, such as screen brightness and screen size, would increase usability and accessibility of these devices.

The majority of participants enjoyed their experience using the mobile-sensing watches and accompanying app, which is consistent with previous research [33,34]. Although some studies have found privacy concerns about personal health data collection and location tracking to be potential barriers to using mobile-monitoring devices among the SMI population [35], qualitative data revealed that these concerns were not present for Veterans in the study.

The convenience and capability of wearable devices to monitor PGHD may be particularly helpful for individuals with SMI, as members of this population often experience psychological barriers to physical activity and other healthy lifestyle choices, which can have negative impacts on their overall well-being [36]. Interviews revealed that mobile-sensing devices can help address this challenge by increasing participants’ levels of motivation to engage in physical activity from week to week upon seeing their PGHD, such as their daily step count, collected by the wearable devices. Individuals also reported that they appreciated the ease and convenience of passive monitoring and that the devices helped increase their understanding of their overall health, suggesting that incorporating these devices into participants’ daily lives is both feasible and beneficial.

The qualitative reports also brought to light some challenges of using mobile-monitoring devices. Some physical aspects of the design of the watches, such as the buttons on the device being difficult to operate, became barriers to use for several of the participants. Individuals also shared that wristbands being uncomfortable led to some challenges keeping the devices on for extended periods of time, especially during sleep, and that the lack of brightness of the watch face screen made it difficult for them to see their data when outside during the day. Additionally, differing comfort with using technology made it particularly difficult for several participants to use the device’s companion cell phone app, which has previously presented challenges of wearable device use for this population [37]. In future research, more detailed training with participants upon initially receiving the wearable device, as well as ongoing technical support during device use, may be beneficial to ensure individuals are able to access all PGHD and amenities provided by these devices. In addition, using peer support coaches to help with mobile health interventions has been shown to be helpful for Veterans with SMI [38], and these active supports may also be beneficial to increasing the accessibility and understanding of mobile-sensing devices for this population.

This study has some potential limitations. In this sample, we interviewed Veterans with SMI receiving treatment at the GLA VA, an urban area. These individuals volunteered to participate in a study involving the use of wearable devices that track their physical activity and location, and therefore, the thoughts and experiences of these participants may differ from those of other individuals who did not consent to participate and use these devices. Secondly, the interview guide was informed by the Technology Acceptance Model, which did not explore differences in participants’ thoughts or experiences based on sex, race, or age, and potential differences in these areas did not come up organically during interviews, which limits our understanding of how these characteristics may influence perceptions of wearable devices for individuals with SMI. Further study of wearable devices that explore the influences of sex, race, and age is needed to expand our understanding of the perceptions and uses of wearable devices in this population. In addition, although access to a smartphone was not a barrier for any participants in our sample, and research has shown smartphone use in individuals with SMI to be as high as about 60% [18], this obstacle may be present for some individuals. Fitbits can be activated through a smartphone, tablet, or computer, and are able to collect PGHD, such as steps, heart rate, and sleep information, without needing further access to additional technology. However, to use their full amenities, such as seeing long-term data trends and using GPS, the device needs to be synced to the Fitbit app via smartphone or tablet. Health care facilities, such as the VA, may be able to reduce potential technological accessibility barriers through programs aimed at reducing this digital divide, such as via the Office of Connected Care. Finally, participants’ duration using the devices varied from periods of two weeks to about a month. Individuals who had been using the watches for the shorter period had less time to engage with the device and accompanying app, which may have led to decreased PGHD collection and allowed less time for participants to become acquainted with using the device or to experience potential challenges.

Research has increasingly turned to the use of technology to help address barriers of PGHD and communication between patients and providers. Using mobile-monitoring devices and PGHD to improve treatment outcomes for individuals with SMI has been gaining traction, especially as more research uncovers its potential to benefit patients in this population. These devices conveniently collect passive sensor data that can be shared with health care organizations to help fill in gaps on important health trends that patients may experience between medical visits, such as stress levels, physical activity, and sleep quality and duration. Through using these devices, patients could potentially self-monitor their activities and identify changes in their behavior that may be associated with a lapse in mental health. In addition, these devices can provide ample information about various health behaviors that could be used to prevent relapse and modify approaches to patient care. This study indicates that wearable devices should be feasible overall for participants with SMI and that individuals within the population are accepting of these devices. While there may be some unique challenges in this population, incorporating wearable devices into patient care could provide an opportunity to enhance treatment outcomes and prevent relapse. With greater emerging technologies that monitor mental and physical health needs, there is a promising future for the use of wearable devices to support this population.