Fitting dynamic models to epidemic outbreaks with quantified uncertainty: A primer for parameter uncertainty, identifiability, and forecasts

Chowell, Gerardo

We can quantify the parameter correlations using our joint empirical distributions of the parameters (denoted by b Q i where i ¼ 1; 2; …; S) which are derived from our bootstrap approach (described in Section 7)), For instance, for our Example # 2 based on fitting the GGM to the first 15 weeks of the Ebola epidemic in Sierra Leone, parameters b r i and b p i were significantly correlated as shown in Fig. 11 . Despite this, the confidence intervals of these parameters display reasonable uncertainty to reliably characterize the parameters. Example #5: Evaluate the correlation of the b r; b p parameters derived from fitting the GGM to the first 15 weeks of the Ebola epidemic in Sierra Leone (See also Example #2; Fig. 11 ). These parameters are significantly correlated (Spearman rho ¼ -0.99; Pvalue < 0.001).