Skip to main content
. 2021 Sep 22;12:5573. doi: 10.1038/s41467-021-25913-9

Fig. 2. Model-inference system fit to data for the three countries and validation using independent datasets.

Fig. 2

The left column shows the model fit to reported weekly case and mortality data for the UK (a), South Africa (c), and Brazil (e). Dots show the weekly number of cases (in blue) and deaths (in red) per 1 million persons; boxes (middle bar = mean; edges = 50% CrIs) and whiskers (95% CrIs) show the corresponding model estimates. Gray shaded boxes indicate the timing of lockdowns or key periods of restricted activity; horizontal arrows indicate the timing of variant identification and vaccination rollout. The right column compares available, independent measurements to corresponding model estimates. Red dots and error bars show measured prevalence over 10 periods of time from the REACT-1 study for the UK (b), cumulative infection rates from two serology studies in South Africa (d), and cumulative infection rates from two nationwide household serosurveys in Brazil (f). Blue lines and surrounding areas show model-estimated mean, 50% (dark) and 95% (light) CrIs. Model estimates (mean, 50% and 95% CrIs) are summarized over n = 100 model-inference runs (500 model replica each, totaling 50,000 model realizations).