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

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.02. 29.20029421 doi: medRxiv preprint proposed model utilizes the strength of the SIR's dynamic system to capture the primary mechanism of the COVID-19 infectious disease, we are able to predict future episodes of the disease spread patterns over a window of 200 days from the last date of data availability. Some turning points of interest are obtained from these forecasting curves as part of the deliverable information, including the predicted time when daily proportion of infected cases becomes smaller than the previous ones and the predicted time when daily proportion of removed cases (i.e. both recovered and dead) becomes larger than that of infected cases, as well as the time when the epidemic ends. Our informatics toolbox provides quantification of uncertainty on the prediction, rather than only point prediction values, which are valuable to see the best versus the worst. The key novel contribution is the incorporation of time-varying quarantine protocols to expand the basic epidemiological model to accommodate changing transmission rates over time in the population. The toolbox can be used by practitioners who have better knowledge of quarantine and better quality data to perform their own analyses. Practitioners can use the toolbox to evaluate different types of quarantine strategies in practice. All summary statistics obtained from the toolbox are of great importance for public health workers and government policy makers to take proper actions on stop spreading of COVID-19.