Fig. 2: Parameter calibration from case studies.
a The symptom onset time distribution. Raw statistics of three reported data sets (up triangle11, down triangle20 and square21) and their union (solid circle) are shown. The red curve gives the estimated distribution under a maximum-likelihood scheme. Grey thin curves are generated with bootstrapping (see the “Methods” section). MLE maximum-likelihood estimation. b The infectiousness function. The data set contains 66 transmission pairs7. Result of the MLE is given by two exponential functions meeting at −0.68 days (red solid line). Also shown are distributions with a constant bridge (dash line), or with a dome cap (dash-dotted line), with slightly lower likelihood values (see Supplementary Section 2.2). Thin grey curves are from bootstrapping for model #1 (see the “Methods” section). c Serial interval statistics outside the Hubei province in China from January 9 to February 13, 202022,23: whole period (solid circles), first two weeks (open squares), and last two weeks (open triangles). The grey dashed line on the right indicates exponential decay at a rate −0.31/day. The red curve is the convolution of the two red curves shown in a and b.
