Fig 1.
The inputs, output and processes of the proposed method.
Fig 2.
Comparison between the synthesized distribution of time intervals between symptom onsets and confirmations (red solid line) and individual distributions of Sichuan, Guangdong, Anhui, Henan, Jiangxi and Zhejiang (gray data points).
Fig 3.
Comparison between the synthesized distribution of time intervals between symptom onsets and confirmations (red circles) and the fitting curve (blue curve) that obeys the translational Weibull distribution (11).
Fig 4.
Effective reproduction numbers for all provinces in mainland China from 10 January 2020 to 21 February 2020.
The results are averaged over 10000 independent runs, and the cyan areas denote the 95% confidence intervals. In each run, the Monte-Carlo sampling method is applied to infer the symptom onsets. The gray shadows emphasize the situations where the epidemic is under control (Rt < 1).
Table 1.
Results for all provinces in mainland China except Tibet and Qinghai, where the confirmed cases are too few to do statistics.
For each province, we show: (i) the number of cumulated confirmed cases by 22 February 2020; (2) the date t* when Rt got below 1; and (iii) the temporal reproduction number during the last week [15 February 2020, 21 February 2020]. The results are averaged over 10000 independent runs.
Table 2.
The means and variances of the five Gamma distributions used in the simulation model.
Fig 5.
The comparison of effective reproduction numbers directly counted based on the simulation results (blue squares) and estimated by the Wallinga-Teunis method (black triangles) and our method (red circles).
The results obtained by the Wallinga-Teunis method and our method are both averaged over 10000 independent runs.
Fig 6.
Comparison between the estimates of effective reproduction numbers by the true and inferred records of symptom onsets.
The solid blue curves and cyan areas respectively denote the average values and 95% confidence intervals obtained by 10000 independent runs according to the inferred data. The red circles represent the results obtained by the true records. The gray shadows emphasize the situations where the epidemic is under control (Rt < 1). The six plots are results for Sichuan, Guangdong, Anhui, Henan, Jiangxi and Zhejiang.
Fig 7.
Illustration of an example of the QR codes to trace the epidemic in mainland China.
This is the one posted in a public bus in Chengdu City. In the bottom, a Chinese character followed by A11345 is the plate number of this bus, and the character is the abbreviation of Sichuan Province.