Individual-based modeling of Covid-19 in local community settings

Individual based modeling of disease transmission (IBM) offers an attractive alternative to population based approaches e.g. (continuous DE), as it allows a detailed account of biological (risk) factors, environment, and behavior. This is particularly relevant in local community settings (hospital, workplace, school, city district or county), where finite population size and host heterogeneity, in terms social interactions and disease progression, make a ‘continuous approach’ impractical. We develop such IBM methodology to simulate Covid-19 outbreaks in local settings, and explore different control-mitigation strategies. Our models feature multiple disease pathways (asymptomatic, mild and severe) typical of Covid-19, as well as heterogeneous host communities with different susceptibility levels and structured social contacts. Individual hosts undergo SEIR disease progression (Susceptible, Exposed - presymptomatic, Infected-symptomatic, Recovered) of variable stage-duration and infectivity. The crucial (S->E) transition is determined by host ‘contact-pool’ on daily basis. Unlike conventional social-contact network (‘one-to-one’ transmission), our setup features ‘many-to-many’ (multigraph) transmission. Two typical IBM examples include (i) hospital, made of interacting healthcare workers (HCW) and patients, (ii) school/college, where students + staff aggregate in classrooms, dorms and engage in other (social, sport) activities. In both cases, we used available data (a hospital in Wuhan, a college in US) to set up and calibrate our models. Different control/mitigation strategies were explored, including symptomatic and asymptomatic testing and isolation, use of PPE (hospital), social distancing, and contact tracing (college). We assessed the efficacy of each intervention, and resources required to prevent or mitigate the outbreak.