4 - (SYM 17) User-centered Design Methods for Digital Interventions: An Application to the Design of an Automated Messaging Intervention for Mental Health Self-management

Background: User-centered design (UCD) approaches are increasingly adopted to ensure that digital interventions can engage end-users to a sufficient extent to change their attitudes and behaviors and ameliorate mental health symptoms. UCD methods are diverse but center on helping end-users provide iterative input on an intervention design, typically moving through stages of needfinding and elicitation, prototyping, and usability testing. To provide an overview of the UCD process, we describe how we engaged young adults with depression and anxiety in the design of a text messaging program for mental health self-management.

Methods: Young adults (ages 18-25) with at least moderate depression or anxiety symptoms (PHQ-9 or GAD-7 ≥ 10) were recruited from self-screening surveys hosted on a mental health advocacy website. Needfinding and elicitation involved an asynchronous discussion forum and individual interviews wherein the research team asked about use of technologies, mental health challenges, self-management approaches, and ideas for how technology can help. Findings informed low-fidelity prototypes of a text messaging program, and feedback was gathered through synchronous workshops on Zoom. Next, in a series of 1-2 week pilot tests, research staff followed scripts to perform the actions of the messaging system (i.e., a “Wizard-of-Oz” approach). Finally, an 8-week fully automated version of the program was developed and engagement and usability were assessed in a field trial.

Results: Needfinding and elicitation activities suggested that text messaging allows for high levels of reach, and that messaging should be interactive and provide variety in experiences. The resulting intervention therefore delivers daily dialogues that support learning self-management strategies from eclectic psychotherapy orientations and that provide diverse types of interactions (e.g., peer stories, prompts, reflection questions). Based on user feedback, the intervention was refined to reduce message volume, clarify language, refine timing, and increase personalization, both through allowing users to actively set preferences and through passive tailoring of content via reinforcement learning. The 8-week program achieved high engagement (70% daily response rate) and a high usability per the System Usability Scale (mean=78).
Conclusions: Involving end-users in a UCD process led to an engaging and usable text messaging intervention that fits young adults' needs. The intervention will next be evaluated in a randomized controlled trial.