According to the World Health Organization, depression is a widespread chronic mental health issue, affecting over 350 million individuals globally [1]. About 16.1 million adults over the age of 18 years in the United States had at least one depressive episode in the past year [2]. Globally, between 2011 and 2030, the impact of depression on the aggregate economic output is estimated to be around US $5.36 trillion [1]. The reduction of costs related to diagnosis, management, and treatment of mental health issues, including depression, is a crucial target for global health system partners [3-6].

Prevalent methods for the management of depression include the use of both pharmacological (medication) and nonpharmacological (psychotherapy) interventions. The use of antidepressant medications is common and cost-effective; however, its clinical efficacy can be impacted by poor patient compliance to the medication regimens, which often have undesirable side effects (eg, weight gain) or take up to 6 weeks of use, upon initiation, to begin making a measurable impact on depressive symptoms [7-10]. Thus, current literature suggests supplementing pharmacological interventions with cognitive behavioral therapy (CBT) as the most effective combination treatment for the management of major depression in most patients [11,12]. Some studies have suggested that CBT, on its own, is as effective as antidepressant medications for depression [8].

Typically, a psychotherapist delivers CBT during in-person therapeutic encounters at regular intervals. During these interactions, the psychotherapist typically assesses the past and current psychological status of the client using validated instruments such as questionnaires. CBT can provide long-term protection against the relapse of depression [13], but at the same time, this is highly dependent on the skills of the psychotherapist delivering the CBT [14]. Training therapists to deliver effective CBT is expensive and can take 2 to 6 years of additional training [15]. The lack of resources to provide such training and the lack of availability of trained professionals are the most critical limiting factors restricting the widespread uptake of this type of therapy [16,17]. Thus, there is much interest in Web-based CBT as it may remove the above-mentioned barriers [18-21].

Globally, health systems need effective and scalable methods that are cost-effective for delivering CBT to patients suffering from depression. Mobile health (mHealth) is rapidly gaining importance and acceptance and thus presents an exciting opportunity for delivering mental health care [22]. Recent research has found that we can attribute a user’s personality to the way they use their mobile phones [23]. This makes us wonder whether monitoring the user’s real-time data and activity [24,25] in a self-directed CBT can be a supplement to the clinical interventions as depressive adults might show unique traits in mobile phone usage. Few researchers have claimed that CBT delivered via an mHealth app is equally valid and also convenient [26,27]. By tracking the patient’s behavior using mHealth apps, health care providers can gain great insights regarding the patient’s condition and their response to psychotherapy [28]. More recently, CBT for patients with incurable cancers showed anxiety and mood improvements [29]. A systematic review by Rathbone et al on the efficacy of mHealth apps for CBT suggested collaboration, which is our starting point with the use of existing validated evidence [30]. As part of our research in this study, we developed a mobile app that included real-time user location, motion, and voice pitch tracking. However, because of the lack of accuracy of sensors across different smartphones, we had to disable those features as we have reported elsewhere [31].

MoodGYM [32] is a popular, evidence-based interactive CBT website designed for the prevention, treatment, and management of depression in young people with internet access. When we conducted the study, MoodGYM was a free website, and we had access to it until March 2018. Currently, it requires an annual subscription fee for access. This website can be accessed using a computer with a standard Web browser, and it utilizes interactive and multimedia features to deliver CBT. MoodGYM requires user registration and collects basic demographics such as age group, gender, locality of residence, and the highest level of education. MoodGYM provides depression assessment scales for evaluating the progress of the user. Figure 1 contains screenshots of the version of MoodGYM used in this study.

We developed a smartphone app called MoodTrainer, inspired by MoodGYM, and incorporated all the modules that were available in the earlier version of MoodGYM. The current version of MoodGYM in its terms of use describes intellectual property rights, similar to the version we used, which allow for noncommercial use. Our adaptation of the questionnaire is from published material, with appropriate references to the original authors. Before beginning the development of our app, we requested permission from the MoodGYM team to develop a mobile app for usability research purposes. We believe that the MoodGYM content was already relevant to young adults in the US context and did not need any contextual adaptation. We developed certain smartphone-specific features such as location tracking, offline storage, reminders and notifications, voice pitch and motion tracking. The voice pitch and motion tracking were not accurate as a measure of anxiety, based on the research team’s tests (see User Interface Development). So, we eventually removed these features.

The CBT aspect of MoodTrainer uses 6 modules, with each module focusing on a specific type of intervention. Previously mentioned evidence from MoodGYM suggests that the sequential format of the modules helps to improve depressive mood [39]. The modules available in the app, in the order of appearance, are as follows:

In addition to these modules, CBT also provides scales to evaluate the user’s depression and anxiety status using Goldberg Depression and Anxiety Scales [40]. These scales are designed to aid general physicians and nonpsychiatrists in detecting depression. These scales have a specificity of 91% and a sensitivity of 86% [40]. The CBT aspect also helps in identifying dysfunctional thoughts with the help of warped thoughts scale. Parslow et al developed this scale and implemented it in MoodGYM [41]. This is a 20-item scale that covers 7 different areas such as the need for approval, love, influence on others, the need to succeed, perfectionism, external requirements for happiness, and expectations of rights. Research suggests that the warped thoughts scores correlate with the levels of depression and anxiety symptoms [41]. In addition to these scales, there are other evidence-based questionnaires integrated into the CBT modules such as (1) the Life Whacks Questionnaire [42], (2) the Measure of Parenting Style [43], and (3) the Pleasant Events Schedule [44].

The study received ethics approval from the institutional review board of Indiana University, with protocol number 1708867512. We used a 2-arm randomized trial design, where the enrolling participants are randomized into 2 groups—the Web-based CBT group or the mobile app–based CBT group.

The mobile app for CBT is targeted for the management of depression in adolescents and young adults. Although it might be beneficial to exactly target clinically depressed adolescents and young adults, it was not feasible for this study, as that would require Food and Drug Administration approval for 510(k), a US market approval rule, as a mobile medical app and would require the trial to be conducted under the supervision of a mental health practitioner. The student population from which we selected our participants matched the target audience age group, including the 18+ years requirement by the MoodGYM instruments. As CBT aims to improve mood and reduce anxiety, we conducted the trial during the week of final exams, which is, as shown in previous studies, usually the time when students experience anxiety for 2 to 4 weeks [45,46]. This also informed our trial study period of 2 weeks [47].

The participants in the study were recruited by convenience sampling. We approached the participants in usual congregation areas such as waiting areas, cafeterias, and designated workstations. During the initial meeting, the participants were provided with the study details and recruited if they fulfilled the inclusion criteria and consented to participate in the study. The inclusion criteria were as follows:

The study participants were randomized into 2 groups, with one group receiving a link to download the app and another group receiving log-in information to the MoodGYM website. Participants were given identifiers that were used as username in MoodGYM and as GA ID in MoodTrainer. There were 15 male and 15 female subjects, with age ranging from 24 to 44 years and a median age of 32 years. All the recruited subjects were graduate students from the health informatics program and had clinical background in pharmacy, nursing, dentistry, or medicine.

We collected data on the usability of MoodTrainer and MoodGYM using a heuristics survey with a 5-point Likert scale for responses based on Nielsen heuristics [48]. The heuristic characteristics which were studied are summarized in Table 1.

We evaluated the feasibility and acceptability of the mobile CBT using a heuristic evaluation survey and evaluated the user engagement through GA in MoodTrainer and modules completed in MoodGYM. At the end of the 2 weeks, participants received an email with the heuristic evaluation that was developed using REDCap [49]. This survey included questions regarding the features of the mobile phone–based and the Web-based CBT and the self-reported usage of the interventions. The participants were also asked to submit the summarized scores of both the depression and anxiety scales present in the app using the settings page available in the app. These data included details about when the form was completed, and scores were reported anonymously to a research-hosted District Health Information Software v2 instance, a popular open-source health management information system.

The MoodTrainer app was designed with a focus on improving user engagement. The app provides location-based feedback and visual feedback of the progress of the user’s CBT module. The progress visualization of MoodTrainer was developed using Chart.js, a JavaScript library for interactive visualization. On the basis of the scores generated by the CBT scales, a chart was generated to indicate the user's progress across different modules.

Providing location feedback is one of our main innovations, a feature never implemented in a CBT app before. This feature captures the location of the user using the native device API at regular time intervals and displays it to the user on a map. Google Maps API was used to generate a map and create markers of the various locations where the user had been in the past 24 hours. On the basis of this, the user will be able to get appropriate feedback if they are spending more time in an isolated place. Initially, we planned to build a motion-sensing feature using the device’s accelerometer to detect agitations or violent movements. However, integrating this feature did not work as anticipated and had to be removed from the final version of the app. MoodTrainer also provides interactive feedback to the users regarding their performance across different CBT modules. These unique features of MoodTrainer ensure that the user is constantly engaged with the app. The level of real-time feedback provided by MoodTrainer is unique and has not been implemented previously in a CBT. The literature suggests that smartphone-specific features were not utilized in any of the previous apps that were designed for the management of depression [50].

The participants were given these applications for 2 weeks to use, and at the end of the 2-week period, the participants were provided with Nielsen heuristic evaluation survey. The results of the survey are as given in Table 2, which summarizes the responses of the heuristic evaluation survey respondents who responded Yes and Fisher exact test results for the responses.

The app usage statistics were generated using GA can be seen in Figure 3. GA provided detailed metrics of app usage, such as the number of sessions logged, screen views, number of screens per session, and session duration. During the study period, compared to 52 in MoodGYM, a total of 67 sessions were logged by the MoodTrainer users, and there were 608 screen views during this period. On an average, the users viewed about 9 screens per session, and the average session duration was 3 min 34 seconds. During the study, it was also observed that the location feedback screen was the second most-viewed screen, with 6.0% (37/608) of the total screen views. This indicates that the study participants were interested and often looked at the location history. The progress page was among the top 5 most viewed pages during the study period, and it had about 3.8% (23/608) of the total views.

As can be seen in Table 3, 25% of the sessions lasted for about 3 min to 10 min, with an average session duration of 6 min. Furthermore, 10% of the sessions lasted between 10 min to 30 min, with an average session duration of 15 min and 24 screens per session. During the study duration, it was found that 16% of the sessions had a screen depth of 3 screens and lasted about 1 min 30 seconds min, and 11% of the sessions had a screen depth of 20+ screens and lasted about 8 min 30 seconds min. It was also observed that out of the 5 modules, on an average, MoodTrainer users completed about 2.5 modules and MoodGYM users completed only 1 module.

This study is among the first to develop a mobile phone–based CBT on the basis of a Web-based CBT and compare the differences in the medium of CBT delivery. We selected MoodGYM as it is widely used in a global context. Its focus on mood disorder allowed us to apply it to the student context, where we have seen mood issues come up often. The results of our limited trial indicate that the MoodTrainer app is more usable when compared with MoodGYM. Previous studies, such as a study by Watts et al [51] in 2013, demonstrated the efficacy of delivering CBT over mobile phones. Our study affirms the computer-based versus the mobile phone–based CBT comparison for the same CBT content, with additional smartphone-specific features. More recent reviews [52] and studies [53] also show that app-based CBT is more usable and engaging. However, it is important to highlight that as more users get comfortable with smartphone interfaces and more accustomed to spending time on devices, this might simply be a reflection of user familiarization. The time spent on the app and the number of sessions logged indicate that MoodTrainer is more accepted by the users compared with MoodGYM. The usability evaluation of the app reveals that it has a user-friendly design. The app can be monotonous after a while and needs to be designed with more colors and better imagery. As we took most of the graphics and content without much change from the Web-based CBT, there is an opportunity to customize the content for delivery through mobile phones in the future. The users of the MoodTrainer app were able to access the location feedback feature. This page was also the most-viewed page during the study period. This might indicate that location tracking was useful and might improve user engagement. The statistical analysis of the survey responses showed that there were no differences in the responses between the MoodGYM and MoodTrainer users, except for a statistically significant difference in the navigation of the application. The mobile app was easier to navigate than the Web application. The session durations which were logged by the users showed that most sessions were between 3 min to 10 min. In terms of self-monitored CBT or note-taking, this may be considered usual time of daily activity. The median age of the subjects was 32 years, with the youngest at 24 years and the oldest at 44 years. So, even though our study was conducted among student population, the sample was not millennials. We also reflected that even though our subject population were not identified as depressed or having mood disorders, referring to them as mentally healthy was also not accurate. Mood disorders are fairly prevalent among college students, and many do not seek help as these are fairly temporary. Thus, we avoid the claim that our subjects are healthy, as we deliberately did not compare their clinical scales as this was only a usability study.

This study was restricted as a usability study and did not evaluate the clinical efficacy of the application because of cost and time constraints. The study included healthy young adults, although including young adults with depression would have given an appropriate understanding of the clinical efficacy. Depressed adults might also have cognitive issues, but there is no clear evidence on this yet. The apps were provided to users only for a brief period of 2 weeks.

We have attempted to make modifications based on user feedback and released the app to Google Play Store. Evaluating the app for clinical efficacy in a depressed population is the next step.

We presented the first evaluation of a mobile phone–based CBT delivery, where we implemented smartphone-specific features. The study also revealed that CBT delivered by this method was more intuitive and user-friendly than Web-based CBT. This method of CBT delivery might thus be more accessible and cost-effective, with improved user engagement. Such apps being more accessible have great potential in limited-resource settings, where access to face-to-face CBT is a challenge. The MoodTrainer app does not aim to replace a clinical care provider but can supplement it as a self-supporting tool. This app might be utilized by clinical psychologists as a continuum of care to prevent the relapse of multiple episodes. It can also be used as a monitoring aid to track a patient’s progress after the therapy or during the therapy.