Fig. G.2.

Derivation of the time-dependent probabilities of detection$p_{1\mathit{Swe}}\left(t\right)$and hospitalization$p_{2\mathit{Swe}}\left(t\right)$, for Sweden.A: prevalence data for Sweden from Folkhalsomyndigheten 1, Folkhalsomyndigheten 2, with three alternative fits: a linear, a Gompertz and an exponential (for comparison). B: number of detected cases over total cases on time, inferred from prevalence, for the three fits to prevalence of panel A. The final functional form for $p_{1\mathit{Swe}}\left(t\right)$ (the probability of being detected if infected) is in green. C: data for daily new detected cases, people entering hospital and people entering ICU (not to be confused with the numbers of people occupying hospitals and ICUs displayed in Fig. 2 of the main text). D: in blue the ratio of people entering hospitals divided by new daily detected cases two days earlier, which we employ to build the piece-wise linear function (in green). This is assumed for $p_{2\mathit{Swe}}\left(t\right)$, i.e. the probability of being hospitalized when detected positive.