How does the outbreak of 2019-nCoV spread in mainland China? A retrospective analysis of the dynamic transmission routes

Xiandeng Jiang; Le Chang; Yanlin Shi

is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.03.01.20029645 doi: medRxiv preprint available data [12] . Rather than employing the SEIR, we develop a time-varying coefficient sparse vector autoregressive (VAR) model. Using the least absolute shrinkage and selection operator (lasso) [39, 40] and the local constant kernel smoothing estimator [41] , our model is capable of estimating the dynamic high-dimensional Granger causality coefficient matrices. This enables the detection and visualization of time-varying inter-province and self-transmission routes of the 2019-nCov on the daily basis. The resulting "road-map" can help policy-markers and public-health officers retrospectively evaluate both the effectiveness and unexpected outcomes of their interventions. Such an evaluation is critical to winning the current battle against 2019-nCoV in China, providing useful experience for other countries facing the emerging threat of this new coronavirus, and saving lives when a new epidemic occurs in the future.