Modern statistical approaches for epidemiological modelling and uncertainty quantification

Thesis

Advances in statistical methodology and computing power are enabling the development of increasingly sophisticated models of infectious disease dynamics, transforming our understanding of the spread and impact of these diseases. These advances allow us to perform more robust inference, make better predictions, extract more information from existing data sources, and harness novel data sources. As exemplified by the COVID-19 pandemic, such progress is crucial for our highly connected world, which faces an increasing number of infectious disease threats. The need for adaptable, reliable, and innovative statistical methods for the analysis of epidemiological data is clear.

This thesis responds to this need by introducing a series of methodological contributions to the field of infectious disease modelling. In Chapter 3, we present a decision-theoretic approach to uncertainty quantification, linking statistical first principles directly to epidemiological applications. In Chapter 4, we extend two popular epidemiological models to demonstrably improve their uncertainty quantification. Later chapters develop novel methodology for fitting epidemic models to data (Chapter 5), for robust modelling of epidemic prevalence survey data (Chapter 6), and for leveraging wastewater sampling data alongside traditional data sources (Chapter 7). Chapter 8 examines risk-related behaviours in England during the COVID-19 pandemic, and compares inferences made from traditional survey data to those derived from novel data sources. Finally, in Chapter 9, we revisit the models introduced in Chapters 4, 5, and 6 through the framework introduced in Chapter 3.

Authors: Steyn, N

• DOI: 10.5287/ora-oreedpnoq
• Type of award: DPhil
• Level of award: Doctoral
• Awarding institution: University of Oxford
• Language: English
• Keywords: uncertainty quantification; COVID-19; epidemiology; infectious disease modelling; statistics
• Subjects: Epidemiology; Statistics