Fig. 5: Schematic flowchart of data and models for this study.

A prior on the basic reproduction number \({R}_{0,v}\) was used for each outbreak, with a hyperprior varying by SARS-CoV-2 lineages (see Supplementary Table 5). Then, we estimated the instantaneous reproduction number (\({R}_{{{{{{\rm{t}}}}}}}\)) based on the observed daily new cases. By comparing observed \({R}_{{{{{{\rm{t}}}}}}}\) with \({R}_{0,v}\) in a Bayesian inference model, we estimated the coefficients of variables to assess their effects on curbing COVID-19. Ten NPIs were divided into four categories: social distancing measures (yellow), polymerase chain reaction screening (red), contact tracing (green), and facial masking (blue). Finally, an Infectious-Intervention-SEIR-Vaccination (ISEIRV) model was built to simulate the timing and intensity of NPI implementation and elimination strategies under diverse transmission scenarios. The prior information for parameter estimation within the ISEIRV model was informed by the effectiveness of each NPI category.