Description

The ability to predict the spread of a disease during a pandemic is of great societal importance. Process models use understanding of the underlying mechanisms, but these have great limitations due to the complexity of interacting processes and incomplete knowledge of how a new virus spreads. A systematic analysis of available data of two main drivers, notably climate and population density, can shed light on this issue. In the topic you will train a Gaussian Regression model to driven by hourly weather reanalysis data to reported WHO data of COVID morbidity and mortality numbers.
This topic is suitable for SE students.

Objectives and Methodology

The objective of this study is to develop and apply a data driven model to explore the relation between seasonal weather patterns and COVID infections. The method includes the following steps: (1) Describe relations between weather variables and the morbidity rates of respiratory diseases reported in the literature; (2) train an existing Machine Learning algorithm to KNMI and RIVM data of The Netherlands (3) Extend the approach to other countries using ERA-5 reanalysis time series and WHO reported COVID data (4) Discuss the potential and limitations of data driven versus process models for infection rate modelling. The possibility exists to use MODIS night time street light observations as additional source of information about population density and welfare.

Further reading