Efficacy of Linear Regression Modelling of SARS-CoV-2 cases based on local wastewater surveillance

Abstract

In the ongoing SARS-CoV-2 pandemic, there is a need for new strategies for surveillance and identification of arising infection waves. Reported cases of new infections based on individual testing are soon deemed inaccurate due to ever changing regulations and limited testing capacity. Wastewater based epidemiology is one promising solution that can be broadly applied with low efforts in comparison to current large-scale testing of individuals. Here, we are combining local wastewater data from the city of Dresden (Germany) along with reported cases and vaccination data from a central database (Robert-Koch-Institute) with virus variant information to investigate the correlation of virus concentrations in the wastewater and reported SARS-CoV-2 cases. In particular, we compared Linear Regression and Machine Learning (ML) models, which are both revealing an existing correlation of virus particles in wastewater and reported cases. Our findings demonstrate that the different virus variants of concern (Alpha, Delta, BA.1, and BA.2) contribute differently over time and parameters vary between variants, as well. By comparing the Linear Regression and ML-based models, we observed that ML can achieve a good fit for training data, but Linear Regression is a more robust tool, especially for new virus variants. We hereby conclude that deriving the rate of new infections from local wastewater by applying Linear Regression may be a robust approximation of tracing the state of the pandemic for practitioners and policy makers alike.