Fig. 1: Under-reporting and delayed reporting noise.
We simulate true infection incidence It over t (in days) from a renewal model (equation (9) with Ebola virus dynamics) with reproduction number Rt that switches from supercritical to subcritical spread due to an intervention. a, Under-reported case curves (50 realizations, various colors) with reporting fractions sampled from the distribution P(ρt) for sample fraction ρt plotted in the inset (mean sample fraction is \(\bar{\rho }\)). b, Delays in case reports (50 realizations, various colors) from the distribution P(δx) for delay δx (in days) plotted in the left-hand inset (mean delay is \(\bar{\delta }\)). We also provide the true Rt as the right-hand inset (red). The main question of this study is how we quantify which of scenarios a or b incurs the larger loss of the information originally available from It, ideally without simulation.
