Novel coronavirus 2019-nCoV: early estimation of epidemiological parameters and epidemic predictions

Jonathan M Read; Jessica RE Bridgen; Derek AT Cummings; Antonia Ho; Chris P Jewell

We estimated the transmission rate and the recovery rate (the inverse of the infectious period). We assumed that the latent period was 4 days, based on an estimate of the incubation period of SARS, a related coronavirus (Lessler et al., 2009) . This is similar to the estimate of 4.4 days from initial characterisation of 2019-nCoV cases. We make the assumption that the latent period approximates to the incubation period. We also estimated the ascertainment rate within Wuhan, and the initial number of human infections present in Wuhan when the market was closed. Confirmed cases in Chinese cities and other countries/regions reported as of 22 January 2020 were used for fitting; from 23 January cases were not reported for Wuhan and other locations within Hubei but only at the aggregate province level. Fitting was achieved by treating the ODE system as representing the mean number of new cases per day in our study period, and assuming that the observed number of new cases were (approximately) Poisson distributed around this mean. Given the model and data, parameter inference was achieved by maximum likelihood estimation using Nelder-Mead optimisation as implemented in the optim() function in the R statistical language (R Core Team 2020); see https://github.com/chrism0dwk/wuhan/tree/v0.2 for R code, case data, and prepared datafiles.