Figure 3. The characterization of the relationship between the viral concentration measured by qPCR and viral culture.

The best fit of using a saturation model (in a form of $\mathit{\text{qCulture}}\left(t\right)=V_m\frac{V(t{)}^n}{V(t{)}^n+V_c^n}$, where $V\left(t\right)$ is the viral load measured by qPCR) to the data from (A) [17], $V_m=6.59$ log10(PFU/ml), $n=3.91,V_C=5.39$ log10(PFUe/ml) and (B) [15], $V_m=6.78$ log10(TCID50/ml), $n=3,V_c=3.64$ log10(PFUe/ml). Black circles are data points. Mean predictions are given by the red curves, and the shaded areas indicate a 95% confidence interval. 12,500 samples are drawn from the population-level posterior distribution to calculate (C) viral shedding period (in days) (defined as the duration when viral load (measured by qPCR or viral culture) is above 10 PFUe/ml or PFU/ml) for different age groups.