Artificially Intelligent Diabetes Assistant (AIDA) for African Americans with Type 2 Diabetes

PROJECT ABSTRACT Diabetes, a chronic disease associated with significant morbidity and mortality, is disproportionately represented in African Americans. Diabetes self-management education and support (DSME/S) is crucial to delaying or preventing disease progression; however, lack of access to proper healthcare due to financial, cultural, knowledge, geographic, or physical barriers places this disadvantaged population at further risk for poor disease outcomes. Mobile health (mHealth) technologies have shown promise in improving health outcomes, but have important limitations for management of this complex disease. For example, they can lack personalization to users? cultural, health, and educational needs. Some technologies, like app and wearables, require high self-motivation, health literacy, and knowledge for successful utilization. Moreover, most technological solutions do not support the patient-provider collaboration critical to sustained behavioral change and ongoing diabetes care over a lifetime. We propose to build an artificially intelligent diabetes assistant (AIDA), a chatbot-powered telehealth platform that is designed to provide highly interactive, automated diabetes self-care support and facilitate remote patient-provider collaboration. Through dynamic, instant-messaging based conversations, AIDA will provide interpretive feedback of blood glucose levels, conduct assessments (e.g., medication adherence, mood), and deliver real-time, culturally sensitive interventions using smartphone technology. AI technology will enable increasingly targeted, individualized interventions. Integration into a larger communication platform will allow providers to remotely assess, monitor, and intervene using real-time patient data collected by the chabot. AIDA will be a low cost, smart-phone based, ecologically relevant tool that is highly innovative, engaging, and impactful. This research will be conducted in collaboration with urban-based community clinics and University of North Carolina researchers specializing in diabetes, health disparities, African American population, telehealth, and behavioral change interventions. In this Phase I SBIR project, we will address the following specific aims: 1) Conduct focus groups and interviews with African American diabetes patients, their caregivers, and diabetes-care providers to collect input on system features, usability design, and messaging content; 2) Develop three AIDA health modules (glucose monitoring, medication adherence, healthy coping) and a virtual health library; and 3) Examine the feasibility and acceptability of the new technology. A future Phase II application will build on Phase I success by creating additional health modules (nutrition, exercise, and risk management) and conducting a clinical trial comparing the efficacy of traditional DSME/S to AIDA support on health and disease outcomes. We envision that AIDA will be marketed to diabetes-care providers, with revenue generated through licensing fees. If found effective, AIDA could be deployed across the national network of diabetes providers/programs, potentially improving health equity in diabetes care.