COVID-19: Forecasting short term hospital needs in France

Clément Massonnaud; Jonathan Roux; Pascal Crépey

We developed a deterministic, age-structured, compartmental, Susceptible-Exposed-Infectious-Removed (SEIR) model, as a set of differential equations ( Figure 3 ). The population was divided into 17 age-groups (5 years age-band), and can be either susceptible (S), exposed to the virus but not infectious (E), infected and infectious (I), or removed from the chains of infection (R). We used an inter-individual contacts matrix for the French population estimated by Prem and colleagues 25 to simulate age-dependent mixing. We considered an incubation period of 5 days and a contagious period of 11 days 19 . We assumed an average length of stay in hospitals and in ICU wards of 15 days for all patients, of all ages. We estimated the age-dependent susceptibility of the population based on the age distribution of infected cases reported by Wu and colleagues in China 11 . We standardized the Chinese age distribution of cases to the French population to estimate the expected age distribution of cases for France. We then fitted the age dependent susceptibility vector using the R implementation of the subplex algorithm provided by the nloptr package.