Social influencers reduce infection burden and modify epidemic lag in group-structured populations

The impact of social media influence on social learning, health behaviour, and population health is a new, rapidly developing field of research. Previous research focused specifically on online social influencers suggests they may be able to affect group-level diffusion of health-protective behaviours such that they modify epidemic outcomes. The impacts of social influence on behaviour and epidemics depend on whether the disease has a low enough basic reproduction number ($R_0$) that social dynamics have time to diffuse. However, formal models have yet to test the intuitive hypothesis that online social influence is sufficient to generate tangible, real-world effects on epidemics. We develop an agent-based model that incorporates social influencers into an epidemic scenario to test hypotheses about how the presence of competing influence messages affects the diffusion of health-protective behaviours throughout a population, and thereby alters the course and outcome of infectious disease epidemics. We find social influencers had a persistent, independent effect on peak infection intensity and total infection burden of the epidemic, with the greatest effects in highly homophilous scenarios. The presence of health-protective influence effectively flattened the epidemic curve despite equal presence of anti-protective influence.