Local vs. Global Risk Perception’s Effect in Dampening Infectious Disease Outbreaks – An Agent-Based Modeling Study

This work explores two key dimensions of the dissemination of information on disease prevalence during an outbreak. First, we point out the necessity of prevalence data to be collected (and communicated) in different scopes, ranging from direct contacts to the whole population. Second, we emphasize the need to minimize the time between the beginning of contagiousness and case reporting, striving for near real-time reporting. To demonstrate the significance of these two dimensions, we introduce an agent-based toy model with an SEIR scheme on a variety of network topologies (random, regular, small-world, and scale-free networks). We compare four different disease mitigation strategies: A base strategy, which allows uncontrolled spread, a global strategy, where susceptible agents reduce their infection probability according to the system-wide prevalence and two local strategies, where susceptible agents reduce their infection probability according to the prevalence among either their first-order or their first-and-second-order neighbors. Across all four network types, we find that the local strategies better reduce both the outbreak’s peak height and total size, with the first-order local strategy proving most effective. However, when a lag between becoming contagious and reporting one’s disease state is introduced, the prevalence information of one’s first order neighbors becomes less useful. In this scenario, the first-and-second-order local strategy performs as good as or better than the first-order local strategy in terms of reducing outbreak size, in all networks except the random network.