Using random testing to manage a safe exit from the COVID-19 lockdown

Markus Mueller; Peter Derlet; Christopher Mudry; Gabriel Aeppli

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.09.20059360 doi: medRxiv preprint FIG. 3. Our algorithm implements policy releases and restrictions aiming at maintaining a vanishing growth rate. It intervenes whenever the estimated slope of the fraction of infected people is found to be non-zero, here with confidence level α = 3. We plot the model infection fraction U (t)/N and the detected infection fraction i(t) as a function of days in panel (a). The model growth rate k(t) (solid line) and the estimated growth rate k est at times of intervention are shown in panel (b) for the parameters i(0) = 0.0012, k 1 = 0.1, and a test rate of r = 15 000 day −1 . The dashed blue line corresponds to a history of interventions where we assumed that the effect of policy interventions is better known (described by an uncertainty parameter b = 0.9, instead of b = 0.5), so that convergence is much faster.