We propose to use statistical methods we have used for Ebola virus disease forecasting in order to project COVID-19 transmission at the state and municipal level, producing testable forecasts. These will feature continuously updated estimates of the reproduction number (number of cases per case) and permit us to assess the benefits of current interventions (social distancing, school closure). We also propose to conduct a close analysis of the fraction of cases traceable to known cases. This statistic can be useful because it can indicate transmission through unknown routes or through asymptomatic cases, but it can also be influenced by the efficacy of contact tracing itself. Many cases that would have otherwise occurred are caught and prevented by contact tracing and isolation. We will conduct network simulations to determine when large values of this presage epidemic growth (depending on the reproduction number, and timeliness and yield of contact investigation). Finally, we will use detailed network simulation to yield a pandemic preparation road map looking into the future. Specifically, specific events (rate of increase of cases, large number of untraceable cases, cases in varying geographic areas) will yield specific actions (school closures, mass gathering abrogation) despite uncertainty in the mode of transmission. These detailed models will also be used for real time assessment of the timing and duration of school closure.
Real-time estimates of transmission potential will permit assessment of effect of control measures (including school closures). We will use network simulation models, permitting inference based on the fraction of cases untraceable to known cases.
