Fig. 4: The relative information in epidemic case data and death counts.
Using \(\theta ({C}_{1}^{\tau })\) we compare the information in case curves \({C}_{1}^{\tau }\) and death counts \({D}_{1}^{\tau }\) under assumptions that lead to equation (8). We examine various case reporting strategies parametrized as Beta distributions with \(\bar{\rho }\) from 0.07 to 0.38 (ref. 18) and compare the resulting \(\theta ({C}_{1}^{\tau })\) against the equivalent from deaths (which reduces to just the infection–fatality ratio, \(\theta ({D}_{1}^{\tau })={{{\rm{ifr}}}}\)). a, \(\theta ({C}_{1}^{\tau })\) for reported case data at different \(\bar{\rho }\) (each color represents a fixed \(\bar{\rho }\)) as compared with ifr (black dashed threshold). The best reporting strategy is in gray. Inset: proportion of case reporting distributions from the main plot for which \(\theta ({C}_{1}^{\tau }) > {{{\rm{ifr}}}}\). b, Those distributions \({\mathbb{P}}({\rho }_{t})\) at the ends of the empirical \(\bar{\rho }\) range with red indicating when \(\theta ({C}_{1}^{\tau }) > {{{\rm{ifr}}}}\).
