Statistical modelling of COVID-19 data: Putting generalized additive models to work

Over the course of COVID-19 pandemic, Generalised Additive Models (GAMs) have been successfully employed on numerous occasions to obtain vital data-driven insights. In this paper we further substantiate success story GAMs, demonstrating their flexibility by focusing three relevant pandemic-related issues. First, examine interdepency among infections in different age groups, concentrating school children. context, derive setting under which parameter estimates are independent (unknown) case-detection ratio, plays an important role surveillance data. Second, model incidence hospitalisations, for data is only available with a temporal delay. We illustrate how correcting reporting delay through nowcasting procedure can be naturally incorporated into GAM framework as offset term. Third, propose multinomial weekly occupancy intensive care units (ICU), where distinguish between number patients, other patients and vacant beds. With these examples, aim showcase practical "off-the-shelf" applicability GAMs gain new insights from real-world