Quantifying the effect of quarantine control in Covid-19 infectious spread using machine learning

Raj Dandekar; George Barbastathis

It is reported that the infected case count stagnated nation-wide in China by the beginning of March (Cyranoski 2020 ) and in South Korea by the end of March (Fisher & Sang-Hun 2020) ; which eventually led to a stagnation in the quarantine interventions employed in these countries. This is in general qualitative agreement with our forecasting results which show a plateau in Q(t) and R t at R t < 1: Figures 3a,b and 7a ,b. In Italy, as of March 20 th , I(t) is appearing to be linear (Figure 4a) , which is consistent with lower rates of infections being actually reported (Horowitz & Kirkpatrick 2020) and can be taken as a precursor to stagnation. It is also consistent with adoption of strict movement restrictions by the government shortly before the March 20 th date. We forecast that, for Italy, R t will drop below 1 and Q(t), R t both will stagnate between mid to end of April 2020 (Figure 5a , b) indicating halting of the spread of infection.