Modeling the interplay between disease spread, behaviors, and disease perception with a data-driven approach

Abstract

Individuals’ perceptions of disease influence their adherence to preventive measures, shaping the dynamics of disease spread. Despite extensive research on the interaction between disease spread, human behaviors, and interventions, few models have incorporated real-world behavioral data on disease perception, limiting their applicability. In this study, we propose an approach to integrate survey data on contact patterns and disease perception into a data-driven compartmental model, by hypothesizing that perceived severity is a determinant of behavioral change. We explore scenarios involving a competition between a COVID-19 wave and a vaccination campaign, where individuals’ behaviors vary based on their perceived severity of the disease. Results indicate that behavioral heterogeneities influenced by perceived severity affect epidemic dynamics, in a way depending on the interplay between two contrasting effects. On the one hand, longer adherence to protective measures by groups with high perceived severity provides greater protection to vulnerable individuals, while premature relaxation of behaviors by low perceived severity groups facilitates virus spread. Differences in behavior across different population groups may impact strongly the epidemiological curves, with a transition from a scenario with two successive epidemic peaks to one with only one (higher) peak and overall more numerous severe outcomes and deaths. The specific modeling choices for how perceived severity modulates behavior parameters do not strongly impact the model’s outcomes. Moreover, the study of several simplified models indicate that the observed phenomenology depends on the combination of data describing age-stratified contact patterns and of the feedback loop between disease perception and behavior, while it is robust with respect to the lack of precise information on the distribution of perceived severity in the population. Sensitivity analyses confirm the robustness of our findings, emphasizing the consistent impact of behavioral heterogeneities across various scenarios. Our study underscores the importance of integrating risk perception into infectious disease transmission models and gives hints on the type of data that further extensive data collection should target to enhance model accuracy and relevance.