Leveraging artificial intelligence and social innovation to reduce disparities in COVID-19 testing among African Americans

Abstract Pandemic fatigue—a phenomenon characterized by a demotivation to follow recommended protective behaviors that emerges over time and is affected by one’s emotions, experiences and perceptions—threatens our ability to end the COVID-19 pandemic. Waning vaccine-induced immunity, breakthrough infections, new variants, and

uncertainty all contribute to pandemic fatigue. These ongoing challenges highlight the importance of sustaining COVID-19 mitigation strategies, including COVID-19 testing, over the long run to achieve pandemic control. While pandemic fatigue is an expected and natural response to a prolonged public health crisis, it compromises

our ability to keep members of underserved and medically and/or socially vulnerable populations safe, including African Americans. Given that complete eradication or elimination are not feasible, scientists and public health officials are focused on control measures to make COVID-19 endemic. To achieve endemic status, we must

identify and address barriers to COVID-19 testing within vulnerable populations, including pandemic fatigue. Moreover, we must advance communication science interventions that enable us to determine how variations in the presentation of messages targeting perceived risk for COVID-19 can be leveraged to increase motivation for

COVID-19 testing behaviors, and employ effective communication strategies to mitigate the impact of exposure to misinformation on testing acceptance and uptake. Guided by the Capability Opportunity Motivation—Behavior and Minority Health and Health Disparities Research Frameworks, this study will leverage participatory research

methods, artificial intelligence, and infrastructure from ongoing community-engaged COVID-19 mitigation research to: 1) Host a design-a-thon to develop deep learning computer animations capable of conveying the importance of COVID-19 testing and promoting its uptake in community settings among African Americans in

NC. 2) Determine whether a deep learning computer animation intervention (vs a control) improves COVID-19 testing uptake using a 1:1 randomized experiment. Study results will identify effective COVID-19 testing promotion messages for African Americans with the potential for generalization to other key populations.