Modelling fatality curves of COVID-19 and the effectiveness of intervention strategies

Giovani L. Vasconcelos; Antônio M. S. Macêdo; Raydonal Ospina; Francisco A. G. Almeida; Gerson C. Duarte-Filho; Inês C. L. Souza

The good performance of the RGM exhibited in Fig. 1 , for the two countries where the COVID-19 pandemic is most advanced, encouraged us to apply the model to other countries at earlier stages of the epidemics. Here, however, care must be taken when estimating the model parameters from small time series, since it is well known that the Richards model [20, 21] and its variants [13] are susceptible to the problem of over fitting, owing to the redundancy of the parameters. This may lead, for example, to estimation of certain parameters that are outside of biologically or otherwise reasonable ranges [20] . For example, when applied to the number of infected cases in an epidemics the parameter α should be constrained within the interval (0, 1). Here we apply the RGM instead to the number of deaths, but we assume that the same constraint should be observed. In other words, fits that return α outside this interval are disregarded as not reliable. The unsuitability of the RGM is particularly serious when the available data does not encompass the inflection point t c [13] . As more data is accumulated, the model is expected to become more accurate. However, if strong measures are implemented towards mitigation or suppression, the dynamics of the epidemics is likely to change, so that different parameter models are required before and after the adoption of the countermeasures; see next section.