Integrating AI nowcasting models with digital communications to reduce RSV infections in rural areas

Project Summary/Abstract Public Health 3.0 places an emphasis on the use of innovative technologies for improved community engagement as well as an increased use of rigorous and timely data in order to deliver relevant health messages (1). This priority also aligns with that National Institute on Nursing Research’s (NINR) research lenses focused on reducing health disparities, advocating for prevention and health promotion, and integrating elements of SDOH to advance innovative new systems (2). Among pediatric populations, respiratory syncytial virus (RSV) poses a serious, yet preventable threat. Educational campaigns known to be effective among rural communities are currently hamstrung by hard-to-reach residents, limited data-driven techniques, and availability of resources (4,17). Grounded in previous work (56-59), Social Cascade enables trusted community messengers (e.g., pediatricians) to share reliable, accurate health-related content on popular social media channels–the #1 source for health information among younger generations regardless of broadband subscription (6,38,70). However, health messaging is most effective when it is tailored to a specific community and timed accordingly (12,13,30). As seen in prior work (56-59), hybrid machine learning techniques and composable diffusion models afford promise for dynamically reconciling masses of data sources. In turn, such algorithms can accurately predict RSV infection rates (also known as nowcasting) for a particular community and ensure the right message is being delivered to the right audience at the right time (61-65). Using both temporal and spatial dimensions, interventions have to be proven successful for promoting behavioral change among rural populations for vaccine uptake (64,66) and infectious disease prevention strategies (36). This Phase I proposal will build and operationalize a tailored Spatio-Temporal Nowcasting Model (STNM) to enhance health communications distributed via social media. The novel integration of these elements holds strong theoretical promise for promoting behavioral change and reducing RSV infection rates in the target rural population of eastern NC. Specifically, the plan seeks to answer 1) can the STNM accurately predict RSV outbreak in real time, and 2) does Social Cascade’s data-driven content and delivery system positively impact behavioral change related to the prevention of RSV infections in the target community? As of April 2024, Social Cascade has acquired 65 customers who have posted more than 6,400 pieces of content reaching more than 500,000 families with a total of 26,935 post engagements (e.g., likes, shares, comments). Parents are enjoying the content with 89.3% of Social Cascade posts earning an engagement rate (number of post engagements as a function of views) above the industry standard of 5%. Furthermore, the underlying framework has far reaching impact potential as it easily extends to additional health priorities and population segments aligning to the NINR’s Strategic Plan (2).