Inferring the COVID-19 infection curve in Italy

Pugliese, Andrea; Sottile, Sara

Aim of this manuscript is to show a simple method to infer the time-course of new COVID-19 infections (the most important information in order to establish the effect of containment strategies) from available aggregated data, such as number of deaths and hospitalizations. The method, that was used for HIV-AIDS and was named `back-calculation', relies on good estimates of the distribution of the delays between infection and the observed events; assuming that the epidemic follows a simple SIR model with a known generation interval, we can then estimate the parameters that define the time-varying contact rate through maximum likelihood. We show the application of the method to data from Italy and several of its region; it is found that $R_0$ had decreased consistently below 1 around March 20, and in the beginning of April it was between 0.5 and 0.8 in the whole Italy and in most regions.