The Ockham's razor applied to COVID-19 model fitting French data

TL;DR

This paper introduces a simple, two-parameter model based on Ockham's razor that effectively fits COVID-19 data across French regions, challenging the necessity of complex models for pandemic forecasting.

Contribution

The paper demonstrates that a minimalistic two-dimensional model can accurately reproduce COVID-19 indicators, questioning the need for more complex models with potential nonidentifiability issues.

Findings

The simple model fits multiple COVID-19 indicators across regions.

Complex models may be unnecessary for accurate pandemic modeling.

The approach encourages debate on model complexity versus effectiveness.

Abstract

This paper presents a data-based simple model for fitting the available data of the Covid-19 pandemic evolution in France. The time series concerning the 13 regions of mainland France have been considered for fitting and validating the model. An extremely simple, two-dimensional model with only two parameters demonstrated to be able to reproduce the time series concerning the number of daily demises caused by Covid-19, the hospitalizations, intensive care and emergency accesses, the daily number of positive test and other indicators, for the different French regions. These results might contribute to stimulate a debate on the suitability of much more complex models for reproducing and forecasting the pandemic evolution since, although relevant from a mechanistic point of view, they could lead to nonidentifiability issues.