An epidemiological forecast model and software assessing interventions on COVID-19 epidemic in China

Peter X Song; Lili Wang; Yiwang Zhou; Jie He; Bin Zhu; Fei Wang; Lu Tang; Marisa Eisenberg

COVID-19 outside Hubei has been so far much less severe. Possible reasons for such low proportions of infection and deaths include (i) discontinuing the traffic connections between Hubei and the provinces, (ii) more timely caution and preventative measures taken, and (iii) a comparatively large population that dilutes the exposed group. When Panel A in Figure 12 is zoomed in, some of the observed proportions (black dots) are deviated from the posterior mean or median of the fitted prevalence albeit they all fall in the 95% credible intervals, as shown by Panel A in Figures 13 and 14 . Since the latent process follows the SIR differential equations, there may be a lack of fit for the SIR model to accommodate a very large and complex population of 1.3 billion people, in which most of the subjects are not at risk. The proposed models should work much better for individual provinces, but we did not perform such analyses. The other epidemiological model with an added quarantine compartment as an absorbing state was fitted via our R function qh.eSIR in the package eSIR. The arguments used in qh.eSIR() are almost identical to those in tvt.eSIR(). Note that if the quarantine rate function is set at constant 0, this model will be reduced to a basic epidemiological SIR model. ### Example 4: Dirac delta function of the quarantine process change_time <-c("01/23/2020","02/04/2020","02/08/2020") phi <-c(0.1,0.4,0.4) res.q <-qh.eSIR (Y,R,begin_str="01/13/2020",death_in_R = 0.4, phi0=phi0,change_time=change_time, casename="Hubei_q",save_files = T,save_mcmc = F, M=5e2,nburnin = 2e2) res.q$plot_infection 14 All rights reserved. No reuse allowed without permission. author/funder, who has granted medRxiv a license to display the preprint in perpetuity.