Modelling the COVID 19 pandemic requires a model... but also data! M

We propose to build a SIR-type model for the Covid-19 data provided by the Johns-Hopkins University. The data available for each country are the daily number of confirmed cases and the daily number of deaths. The model is adapted in order to fit the data at an aggregated level like a country. In other words, the parameters of the model change from country to country to reflect differences in dynamics. In particular, the model integrates a time-dependent transmission rate, whose variations can be thought to be related to the public health measures taken by the country in question. A piecewise linear model is used for the transmission rate to take into account these possible variations. The proposed model may seem simple, but it should be understood that it does not pretend to describe the spread of the pandemic in a precise and detailed manner. Its role is to adjust the available data: its complexity is therefore adjusted to the amount of information available in the data. Indeed very few parameters are needed to properly describe the outcome of interest, and the prediction proves stable over time. The model, the parameter estimation algorithm, the method for model selection as well as several plotting routines have been implemented in an interactive, easy to use, web application that allows to visualize the data and the fitted model for several countries (http://shiny.webpopix.org/covidix/app2/). The data used in this application are updated frequently in order to be able to follow on a day-to-day basis what the model predicts for several countries.