Flattening the curve for the Coronavirus

In order to better understand the development of the Coronavirus pandemic, I decided to do my own statistical analysis. I only considered the number of deaths in various countries, as this number is probably more reliable than the number of infections. I then made the assumption that this number grows exponentially, but that the coefficient of growth changes linearly with time:

Simple model for the number of deaths N(t). The exponent α(t) is assumed to change linearly with time.

It is now possible to make a simple fit for each country and to estimate when the curve will flatten and the number of deaths will stop increasing.

I was surprised how well it worked. Below is the result for Italy

Number of deaths from the Coronavirus in Italy

The analogous plot for Switzerland looks as follows

Number of deaths from the Coronavirus in Switzerland

Unfortunately, the United States is not even close to flattening the curve

Number of deaths from the Coronavirus in the United States

It is now possible to compare different countries both in terms of actual deaths

or in terms of the fitted curves

The curves tell us an important message: If strict measures are implemented to stop the spread of the virus, this actually helps. It is the only thing we can do at the moment.

As public service, I have made the Python script for the analysis available. It automatically downloads the data from the Center for Humanitarian Data, performs the analysis and plots the data.

Here is a scientific study of the efficacy of different measures from Imperial College: Estimating the number of infections and the impact of nonpharmaceutical interventions on COVID-19 in 11 European countries.

Here is a useful tool for a more detailed analysis of the disease from the University of Basel: https://neherlab.org/covid19/

BTW, there is a good reason be be wary of exponential growth, as this delightful video shows:

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.