Robust Statistical Estimation of Epidemic Transmission Models on 2019-nCoV Data

Summary

The recent outbreak of a novel coronavirus (2019-nCoV) has quickly evolved into a global health crisis. This project adopts robust statistical methods to estimate key parameters of the transmission model of 2019-nCoV. SEIR models are fit on a group of 84 major China cities and Wuhan connected by traffic networks of large volume immediately prior to Wuhan lockdown. Model fitting is managed using the random sample consensus (RANSAC) algorithm.

The robust estimation enables us to identify two clusters of transmission models, both are of substantial concern,

in the initial stage, the basic reproductive number R0 may be ranging in 8 ~ 14, comparable to that of measles
there has been a large unidentified group of infections (I0) dated back to 1 Jan 2020

Previous model estimation based on maximum likelihood can be severely biased due to the large volume of data reported in the initial stage in Wuhan, subject to delay/misdiagnosis.

Comparision of the estimation results from the consensus analysis and what was believed:

Installation

All procedures are included in a singal Python Notebook, all data up-to-date are provided in data/ folder. Jupyter Notebook is needed to run the project. Alternatively, this project has been made working in cloud-based computational environment (there is a data downloading section following the environment preparation in the notebook). It has been tested on Colab by Google Research, which can be setup by a single click on the "Open in Colab" link at the top of the page displaying NB_RANSAC_Estimation_SEIR_2019-nCoV.ipynb.

Application on Own/Local Data of Infections/Traffic

Such adoptation is straightforward by replacing the data files in data/ folder. More instructions maybe added here. For emergent needs, raise an issue or contact the author.

Updates and Errata

As the project is on an on-going public health event, I choose to release the research on real time. In errata_updates/ folder there are errata and updates to the arXiv article (see Reference below).

fig1a_with_circles.pdf to Fig. 1a, add annotations of the parameter clusters mentioned discussed in text body

NB: this is not updates or corrections to the project implementation, which IS realtime on Github.

Reference

If you use the project in your report/research, please cite the work as follows

@article{Li2020,
    author={Jun Li},
    title={{A Robust Stochastic Method of Estimating the Transmission Potential of 2019-nCoV}},
    journal={arXiv},
    year={2020}
}