Using hospital data for monitoring the dynamics of COVID-19 in France

Abstract

The objective of this article is to show how daily hospital data can be used to monitor the evolution of the COVID-19 epidemic in France.
A piecewise defined dynamic model allows to fit very well the available hospital admission, death and discharge data.
The change-points detected correspond to moments when the dynamics of the epidemic changed abruptly. It is therefore a surveillance tool, not a forecasting tool.
In other words, it can be used effectively to warn of a restart of epidemic activity, but it is not designed to assess the impact of a new lockdown or the emergence of a new variant.
The model, data and fits are implemented in an interactive web application.