Anticipating the novel coronavirus disease (COVID-19) pandemic

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

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04.08.20057430 doi: medRxiv preprint rate observed in our original data sets by investigating the indicators in the surrogate time-series (see Materials and Methods). The surrogate time-series is generated to follow similar distribution (mean and variance) of the data time-series before the episode of a sudden rise in the number, denoted by shaded regions in Fig. 1 (see Materials and Methods). Figure 4 depicts the distribution of the test statistic of the surrogate time-series. Solid lines show the trend estimate obtained for the original timeseries. We calculate the probability of randomness of our observed estimates as the fraction of 1000 surrogate time-series having trend statistic of same or higher absolute values than the original trend, i.e., P (τ * ≤ τ ). The probability to, by chance, obtain similar trend statistic varies from country to country, depicting most significant estimates for Singapore (Figs. 4D and 4L) and UK (Figs. 4G and 4O ). The probability estimates P obtained by bootstrapping the data sets for each of the countries are given in the Table II . While in the case of the US, the probability of randomness in our observed estimates is relatively high (Figs. 4C and 4K), rapid spreading in the epidemic makes it keystone to consider applicability of EWSs to warn-off such events. Overall, we find a low probability of randomness in both the ACF(1) and the return rate estimates for all the cases. However, the observations are more significant for the return rate. This analysis suggests the robustness of the return rate as an EWS in predicting the signals of CSD.