User-Centered Development of a Chatbot for Diverse Adolescents at High Risk for Eating Disorders

Adolescence is the peak onset period for eating disorders (EDs), making ED prevention critical during this time. Comorbid depression and anxiety can put adolescents at further risk for EDs. Chatbots can be a scalable solution, yet minimal research has explored this tool in diverse adolescents for ED prevention and addressing comorbidities. This study utilized user-centered design to tailor a rule-based chatbot to diverse adolescents.

Fifteen adolescents in the USA aged 14-17 years who owned a smartphone screened at risk for an ED (high weight and shape concerns) based on an online screening tool. Participants completed a 30- to 60-min semi-structured interview on Zoom where they viewed the chatbot app's layout and chatbot scripts via screen share and provided feedback in think-aloud format. Interview transcripts were coded using an inductive approach.

Participants' (60% minoritized sexual identity: lesbian/gay, bisexual, pansexual, aromantic/asexual, queer, or questioning; 20% minoritized gender identity; 80% non-White) feedback converged on three themes: "User Experience with Chatbot Structure and Content," "Barriers and Facilitators to App Use and Engagement," and "Relatability of Content to Target Adolescent Population." Participants appreciated the chatbots' interface, stepwise psychoeducation, and relatability of content. Participants noted time constraints and motivational barriers to use, as well as the strict schedule and binge-eating focus within the regular eating content.

Overall, the chatbot resonated with participants and showed potential for use with adolescents at risk of developing EDs once content is adapted to their feedback. The next stage of our work will integrate feedback and investigate the effectiveness of this intervention.