Short-term prediction and impacts of different socio-factors and mobility levels on the spread of COVID-19

Abstract

This Ph.D. thesis started in 2020, during the second wave of COVID-19 in Belgium. Many questions about COVID-19 were unanswered at that time, and this thesis is inspired by several of the questions that were asked by the local and national governments in Belgium. In any infectious disease outbreak, it is crucial to comprehend what impact the disease’s spread to tailor interventions for optimal results. Additionally, having timely predictions about hospital capacity exceedance is also essential to enhance resource preparedness. We aim to propose statistical methods that exploit data at various scales to address these two goals for the COVID-19 pandemic, which can be applied for other possible outbreaks in the future. The significant contributions of our work include an innovative joint model that can estimate the desired parameters without, or with limited necessary individual level information in Chapter 2, a novel use of telecom-based mobility data for the spatial connectivity in Chapter 3 and 4, a pioneering model that can identify areas at a higher risk while accounting for the disease spread in the larger neighborhood in Chapter 5, and an assessment of a gap when using the popular conditional autoregressive (CAR) model in Bayesian disease mapping with fine-scale data in Chapter 6.