Back

Modeling COVID-19 in different countries as sequences of SI waves

Janssen, R.; Mimkes, J.

2022-11-22 epidemiology
10.1101/2022.11.22.22282631 medRxiv
Show abstract

The COVID-19 pandemic has been a huge challenge worldwide for many institutions, researchers, national health organizations, and the pharmaceutical industry. As natural scientists and engineers, we attempted to contribute by calculating models and analyzing data to keep track of the pandemic. While a frequent goal is to predict the next pandemic wave by considering all influencing parameters, we examined methods to calculate a model course of the entire pandemic. This is done by reconstructing the course of infections into multiple model waves that sum up into a pandemic model that is close to the real course. The model wave parameters are varied by an algorithm, such as the Excel solver, to minimize the difference between the real and model courses. By reconstructing the course of infections using the commonly known SIR model, we found that the calculated model parameters were ambiguous and difficult to interpret. In contrast, we found that sequenced SI model waves provide an astonishing precise digital representation of the pandemic course. Until November 2022, we found between six and 16 waves (depending on the country) in each of the 14 countries investigated. The calculated parameters are easy to interpret and are comparable between different waves and countries. These wave parameters may be correlated with the virus types and measures in each country by other researchers. New waves are detectable early as they show a certain deviation from the actual model wave. After the maximum of the last real wave, the model indicates the further procedure for the pandemic course.

Matching journals

The top 6 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.