Key points are not available for this paper at this time.
BACKGROUND: Early treatment is critical to improve eating disorder prognosis. Single session interventions have been proposed as a strategy to provide short term support to people on waitlists for eating disorder treatment, however, it is not always possible to access this early intervention. Conversational artificial intelligence agents or "chatbots" reflect a unique opportunity to attempt to fill this gap in service provision. The aim of this research was to co-design a novel chatbot capable of delivering a single session intervention for adults on the waitlist for eating disorder treatment across the diagnostic spectrum and ascertain its preliminary acceptability and feasibility. METHODS: A Double Diamond co-design approach was employed which included four phases: discover, define, develop, and deliver. There were 17 participants in total in Australia; ten adults with a lived experience of an eating disorder and seven registered psychologists working in the field of eating disorders, who participated in online interviews and workshops. Thematic and content analyses were undertaken with interview/workshop transcriptions with findings from the previous phase informing the ideas and development of the next phase. A final prototype of a single session intervention chatbot was presented to the participants in the deliver phase. RESULTS: Thematic and content analyses identified four main themes that were present across the four phases of interviews/workshops: conversational tone, safety and risk management, user journey and session structure, and content. CONCLUSIONS: Overall, the feedback on the single session intervention chatbot was positive throughout the Double Diamond process from both people with a lived experience of an eating disorder and psychologists. Incorporating the feedback across the four themes and four co-design phases allowed for refinement of the chatbot. Further research is required to evaluate the chatbot's efficacy in early treatment settings.
Sharp et al. (Tue,) studied this question.