Anticipating the novel coronavirus disease (COVID-19) pandemic

Taranjot Kaur; Sukanta Sarkar; Sourangsu Chowdhury; Sudipta Kumar Sinha; Mohit Kumar Jolly; Partha Sharathi Dutta

To estimate statistical indicators anticipating the upcoming shifts in the growth curve, we extract the cumulative daily number of COVID-19 cases up to 35-40 days from the beginning of the epidemic (shaded regions in Fig. 1 ), for each country (note that the EWSs analyses on the fraction of reported case datasets result in qualitatively similar outcomes). To examine whether the system slows down to recover from perturbation while approaching the transition, we estimate the changes in the return rate and autocorrelation at first lag (ACF(1)) of each extracted data for all the nine countries (see Methods). Critical slowing down is reflected in systems near a critical transition through an increase in the autocorrelation. We observe that after nearly 40 days of the onset of the epidemic, the short term memory of the time-series data exhibits an increasing trend in most of the countries (Fig. 2) . However, there are no significant signals of CSD exhibited by ACF(1) for the datasets of India as well as Italy (Figs. 2J and 2P). The return rates for all, except India ( Fig. 2A) , decreases, thus signaling to ex-pect a sudden rise in the number of the COVID-19 cases for these countries. Furthermore, the strength of signals varies amongst countries depending upon the data sets determining the fraction of affected populations in individual countries. For instance, the trends in the UK are observed to be very strong, with ACF(1) approaching close to 1 (see Fig. 2Q ) [21] . Since the time lag of up to almost two weeks is expected for the detection of symptomatic cases, the observed signals of CSD around 40 days indicate that the total cases gathered till then must be infected with the disease around two weeks ago. Thus, suitable preventive and surveillance strategies adopted in the initial 20-25 days are capable of suppressing the COVID-19 outbreak [26] .