The purpose of this iterative usability study was to inform design of a mobile application, MoodRing, for providing feedback to depressed adolescents about mood through the use of machine learning predictions based on smartphone passive sensor data. MoodRing aims to improve self-management for adolescents, increase parental awareness about worsened mood, and provide clinician feedback that is technologically feasible, usable, appealing, and capable of delivering high-quality, actionable data.
An iterative series of focus groups and interviews were conducted with adolescents who had history of depression, parents who had a child with depression, and healthcare providers including care managers, primary care providers, and mental health clinicians. Three focus groups were conducted with each group to introduce prototype mockups, assess ease of understanding data displays, acceptability of passive sensing, privacy concerns. Then secondary prototypes were developed and interviews conducted with 4 providers, 5 parents, and 6 adolescents. Each was presented with a video scenario of how MoodRing may be introduced to them in primary care. Afterwards each was asked to use a think aloud procedures to describe their interaction with the prototype. Adolescents went through 3 scenarios: onboarding, reviewing real data, and learning a coping skill; parents went through onboarding, viewing child’s data, and learning a communication skill. Providers went through viewing the provider dashboard with detailed patient-level data, and programming a coping skill to set up personalized notifications. Participants completed follow-up interview about their impressions, difficulties, feedback. Interviews were transcribed and coded using thematic analysis, as well as based on six overarching themes, based on 4 categories, namely benefits, concerns, suggestions, and conclusions.
Primary care managers understood potential utility of MoodRing but felt it would be a steep learning curve due to general discomfort with mental health triage. Parents respected adolescents would want confidentiality concerns, while both adolescents and parents agreed clinicians should see as much data as possible. In prototype testing, adolescents and parents gave positive feedback about usability overall and specific feedback about visual design iterations, and all recommended personalization. Being able to control privacy settings and manage notifications was thought essential for users to remain engaged. Trends presented in data and graphs was generally grasped. In terms of privacy, users believed the app clearly addressed all privacy concerns as it listed precisely what would be passively collected. One parent described, "I think having the alert would have helped me to intervene sooner than I did. Cuz I didn’t realize that she was feeling the way she did until it was, you know, too late.”
Overall, feedback obtained from focus groups and interviews provided a substantial trajectory for the development of the MoodRing platform. Adolescents, parents, and providers felt that an app collecting passive data from them would be useful as long as it allowed for awareness of how data was being used, respect for privacy, and personalization. Future steps include finalizing the development of the application and testing it in a clinical trials for effect on improving the quality of depression management.
Sources of Support
R44 MH122067 01 NIMH.
© 2022 Published by Elsevier Inc.