Fig. 2: Predicted risk of COVID-19 symptom onset across 347 Chinese cities.

a–d The predicted onset risk before the date of Wuhan lockdown. e–m The predicted onset risk after the date of Wuhan lockdown under two scenarios, i.e., with (e–g, k–m) and without (h–j) Wuhan lockdown. The predicted COVID-19 symptoms onset risk were generated by the extended WKDE model, by using historical confirmed cases and inter-city human mobility data. The historical confirmed cases data included the locations at which these cases had a period of study prior to the diagnosis. The predictions under two scenarios were made by differing the human mobility intensity from Wuhan to other cities after the date of the Wuhan lockdown. Under the lockdown scenario, the outward human flows from Wuhan were regarded as not increasing the onset risk of other cities, due to the dramatic decrease in human mobility intensity, together with strict quarantine measures. Under the non-lockdown scenario, the human mobility intensity was estimated based on that of the corresponding time period in 2019 (see Methods section for details).