Modeling COVID-19 latent prevalence to assess a public health intervention at a state and regional scale

Philip J. Turk; Shih-Hsiung Chou; Marc A. Kowalkowski; Pooja P. Palmer; Jennifer S. Priem; Melanie D. Spencer; Yhenneko J. Taylor; Andrew D. McWilliams

While there is a plethora of models that estimate the impact of COVID-19 in the US, there are far fewer that give localized projections. We note that our mid-May date for the peak infection curve is roughly 3-to-4 weeks later than the projection from the often-cited model from The Institute for Health Metrics and Evaluation [34] . The later uses a Bayesian generalized nonlinear mixed model to examine cumulative death rates and assumes a strict social distancing policy is in place. Using data up until March 13, Columbia University reported a mid-May peak time for NC under no control measures and a start of July peak time under some control measures [35] . The authors caution that their metapopulation SEIR model is designed to capture national trends, and local projections should be viewed as broad estimates. Other models, such as the CHIME model from the University of Pennsylvania Health System, relied on data from three Pennsylvania hospitals to estimate hospital capacity and clinical demand and was not designed to capture changing regional mitigation strategies [36] .