Optimizing vaccination strategies under uncertainty to prevent epidemics

To be prepared for future infectious disease epidemics, this paper considers a data-driven approach for determining optimal vaccination strategies for multi-community settings with heterogeneous populations under uncertain social mixing, disease transmission, and vaccine efficacy. Specifically, we derive an integrated chance constraints stochastic programming (ICC-SP) disease spread model for optimal vaccination strategies to prevent epidemics by keeping the uncertain post-vaccination reproduction number below one at a specified level of risk. A vaccination strategy specifies the proportion of individuals in a given household-type and age-group to vaccinate and the intervention level needed to bound the expected excess of the reproduction number above one by an acceptable level of reliability (or risk). The ICC-SP model is data-driven and uses readily available data on census demographics, age-related disease susceptibility and infectivity, virus variants, and vaccine efficacy defined in three ways: vaccination in terms of effectiveness against infection, symptomatic cases, and hospitalization. This data-driven approach incorporates the decision-maker's level of risk to enable public health policy what-if analyses for future epidemics. A case study using the ICC-SP approach based on COVID-19 data was conducted, and the results of the study provide several insights. The study shows, for example, that to control disease outbreaks vaccination strategies must be combined with a specific intervention level. The proportion of the population to vaccinate to prevent epidemics depends on the criterion of vaccine efficacy used and decreases with increasing intervention level. The study also shows that optimal vaccination strategies prioritize the vaccination of specific households and age groups with high combined levels of relative susceptibility and infectivity.