Evaluations of three computational models (AR, GT, and ARGO) across five outbreaks, based on correlation (r, Panel A), root-mean-square error (RMSE, Panel B), and relative root-mean-square error (rRMSE, Panel C).

The result of the best-performing model for each prediction scenario and metric is bolded. It is important to note that the units of the error (RMSE) are different given that the magnitude of each outbreak was different. The relative error, however, is comparable across outbreaks.