Problem description Mental illness among working adults is of huge detriment to society on a large scale. Not only does mental illness lead to a lower quality of life and negative outcomes for the individual, but it places a burden on the economy, resulting in lost productivity, lost workdays and reduced economic growth. According to the National Institute of Mental Health an estimated 17.3 million adults in the United States had at least one major depressive episode in 2017. This number represented 7.1% of all U.S. adults. Furthermore, according to ADAA anxiety is one of the most common mental illnesses in America with around 40 million adults in the US having anxiety disorders every year. In addition to the millions lost on medical costs from mental illness, studies show an added societal burden. On average depressive symptoms can lead to close to 27 lost workdays per year and 18 days of lost productivity.
While mental illness negatively affects adults in significant ways, primary care doctors identify less than one-third of all depression cases and up to 40% of persons with serious mental illnesses do not receive care or stop intervention programs due to negative stigma. With only one in four adults disclosing their anxiety disorder in time, a much smaller proportion are ever diagnosed and receive treatment for major depression. Lastly, but perhaps most importantly, according to the Center for Workplace Mental Health, 80 percent of adults treated for mental health problems report improvements in their overall satisfaction and productivity.
Study model This study uses is a mental health-focused application that utilizes artificial intelligence to identify symptoms and provides targeted, tailored in-person therapy solutions for its users. The application is meant to augment in-person psychotherapy utilizing online journaling and daily patient assessments to provide curated content and feedback. In brief, the application utilizes machine learning and artificial intelligence to select content (e.g., self-help articles) and provide feedback (e.g., weekly summary score of mood and anxiety) based on user input (e.g., journal entries, daily mood rating)
Study Design The proposed study is a 5-week feasibility study to assess the usability and potential short-term benefits of the mental health platform in a population of adults with established mental health care.
Potential participants will be informed of the study using study brochures provided to their established psychotherapists. All written materials are submitted for approval. Interested patients will be directed to a secure online survey, which will screen them for eligibility. Patients that meet all inclusion criteria and no exclusion criteria will be offered entrance into the study. Of note, no medical records of the patients will be accessed as part of the study.
After the initial screening survey, 45 eligible patients will complete the consent process. The consented participants will be randomized to one of the two study arms: (1) the intervention arm or (2) the waitlist control arm in a 2:1 ratio. The intervention arm is described in detail below. The participants in the waitlist control arm will serve as controls for the study. They will complete the pre-pilot assessment and the post-pilot assessment. It will be made clear that the participants can continue to seek a standard of care services. During their time on the waitlist, participants may reach out to study personnel if they need assistance with their psychiatric care. We aim to have 22 participants complete the intervention arm and in the setting of participant attrition from the intervention arm, participants in the waitlist control arm will be offered entrance to the intervention arm. If recruited off the waitlist, the participant will complete consent, this time selecting "Intervention" as the study arm and will be entered into the five-week intervention study.
Participants in the intervention arm will be sent an individualized redemption code to download the mobile app application on to their personal mobile phones. The participants enter individual login details. During the first week of the study, each participant will be prompted by the application to complete key surveys and assessments at predetermined time intervals. The mental health platform generates a summary health metric on-demand that is a combination of responses entered from the in-app assessments.
A summary health metric will be pushed in-app after the first week of use. Details for scheduling the first in-person therapy session will be provided at the same time as the summary health metric. Participants will be informed that the general health metrics will be shared with the therapist to help the therapist better plan the sessions. The therapy session will be scheduled in a private room on-campus. The therapists will offer standard-of-care psychotherapy (e.g., supportive psychotherapy, cognitive-behavioral therapy). The participants use the app for a total of 5 weeks and receive weekly in-person therapy for a total of 4 weeks (one-week application only lead-in, four weeks application plus in-person psychotherapy).
Two functions of the application that will be independently examined for usage and utility are its journaling function and curated insights. Participants will be asked to use the journaling function daily (more details below). The curated insights are selected by the application based on participant responses to the daily mood and anxiety surveys, as well as built-in sentiment analysis of the journaling.
At the end of 5 weeks, all participants in both study arms will complete the post-pilot assessment using the same assessments as pre-pilot.
Participants who agree to be contacted post-pilot study will be given the option at 3, 6, and 12 months to take the same online assessments as at pre- and post-pilot. This evaluation will be used to test the potential long-term impact of the app. In addition, post-pilot focus groups may be completed. There would be one focus group with therapists and separate focus groups with participants. These optional assessments will be pending study results, funding, and resources and would be submitted in the future as a change in research to the current Institutional Review Board (IRB).
Details of Study Assessments Scales / Questionnaires Mood / Anxiety Meter:
The application will send timed notifications three times a day (morning, afternoon and evening) for the user to pick current mood and anxiety on a Likert scale General Anxiety Disorder-7 (GAD-7) anxiety scale. The app will prompt the user to fill in this assessment every 2 weeks. The GAD-7 parallels the 7 diagnostic symptom criteria that define DSM-IV GAD. The format and temporal framework of the items also correspond to the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) criteria and will facilitate the follow-up review of symptoms and diagnostic processes. At only 7 items, the GAD-7 is substantially shorter than most anxiety screening measures. This instrument has also demonstrated acceptability among non-psychiatric patients and among busy primary care providers and is a valid and reliable screening tool for anxiety.
Patient Health Questionnaire-9 (PHQ-9) depression scale. The app will prompt the user to fill in this assessment every 2 weeks. The PHQ-9 parallels the 9 diagnostic symptom criteria that define DSM-IV. The format and temporal framework of the items also correspond to the DSM-IV criteria and will facilitate the follow-up review of symptoms and diagnostic processes. At only 9 items, the PHQ-9 is substantially shorter than most depression screening measures. This instrument has also demonstrated acceptability among non-psychiatric patients and among busy primary care providers and is a valid and reliable screening tool for major depression.
Mobile Application Rating Scale (MARS): The MARS is a well-established framework for classifying and assessing the objective and subjective quality of apps, as well as the app's perceived impact. It is designed to score apps on the criteria of engagement, functionality, aesthetics, and information quality, as well as app subjective quality. Each MARS item used a 5-point scale (1-Inadequate, 2-Poor, 3-Acceptable, 4-Good, 5-Excellent). The MARS is calculated as mean scores of the engagement, functionality, aesthetics, and information quality objective sub-scales, and an overall mean app quality total score.
Other Online Journaling: The online journaling feature allows the user to enter his/her feelings and thoughts, like in the case of a diary. The user is asked daily "How has your day been? (please provide at least four sentences to describe how you are doing and feeling today)". Every two weeks the user is asked "How have you been the last two weeks? (please provide at least four sentences describing your emotional well-being over the last two weeks)". This second prompt is meant to allow comparison with the PHQ-9 and GAD-7, both of which ask about a two-week time frame's neural network is built and calibrated to provide an emotional context through pre-labeled datasets that tag words, phrases, and sentences to a specific emotion (e.g., fear, sadness, anxiety). The neural network model then analyzes the input from the journaling features and assigns it an emotion weighting. User input is then analyzed and the output indicates what condition participants are likely currently experiencing based on participant's entry.
