DARA: Development of a Chatbot Support System for an Anxiety Reduction Digital Intervention

This paper presents the development and preliminary evaluation of a chatbot, “DARA,” which is designed to be added to a digital mental health (DMH) cognitive bias intervention targeting anxiety. Although “supportive accountability” (the addition of human coaching to a digital intervention) may help reduce attrition, human support is not always well-suited for DMH interventions targeting anxiety or emotional difficulties (due to the inherent challenge of social interaction among individuals in these populations.) In response, a virtual conversational agent was designed after evaluating support needs with respect to embodiment/non-embodiment, free-text response/quick replies, and domain-free/domain-specific conversational properties. DARA's usability and comprehensiveness were evaluated among 12 subject-matter experts using a suite of quantitative and qualitative usability questions, including the PSSUQ Version 3. Usage data were also recorded from chatbot interactions. Results suggest that chatbots may be particularly appropriate for anxious users, although navigation and lack of depth in chatbot responses may pose concerns. We suggest that these concerns can be resolved by technical improvements and hybridization of the human-chatbot support system. We make future recommendations for the design and integration of the DARA chatbot and the use of chatbots in other DMH interventions targeting anxiety.