Cade Rhone
Research Mentor(s): Sun Young Park
Mentor Department: University of Michigan
Authors: Cade Rhone, Eric Junhan Kim, Sun Young Park
Session: Session 6 (3:00pm – 3:50pm)
Presentation Type: Poster 48
Abstract
Black Americans have a higher prevalence of diabetes, particularly type ll, based on genetic, environmental, and socioeconomic factors. Previous healthcare studies discovered that health interventions culturally tailored to Black Americans led to higher trust towards the health interventions for Black patients with chronic conditions, despite their underlying medical mistrust. Building on this literature, two pilot studies were conducted during the COVID-19 pandemic to understand how we can develop culturally relevant chatbots for the healthcare of Black Americans with chronic conditions. The studies found that Black patients expected the chatbots to be more empathetic towards their culture and understand the underlying history of medical mistrust, which informed the design decisions of our current study. Based on these design implications, in this study, we aimed to develop a culturally tailored chatbot for Black Americans with diabetes using GPT API. Through this, we expect to understand whether a culturally relevant chatbot could increase Black Americans’ knowledge of chronic conditions, health literacy, and trust in chat agents. We created two types of chatbots: culturally relevant (CR) chatbots and non-culturally relevant (NCR) chatbots, which differ in terms of resources provided and conversation method. For instance, the CR chatbot provides articles and resources specifically related to Black Americans’ experience with diabetes and motivational messages from other Black patients. On the contrary, the NCR chatbot provides related articles and messages that do not mention race. 20 Black participants with type 2 diabetes and diagnosed after 2020 are being recruited for this study. They are randomly allocated to either of the groups as a between-subject design. Participants receive the link to the website where we deployed the two chatbots for participants to interact with for 2 weeks. Currently, 7 participants have been enrolled and are conducting the study.
