Spatio-temporal endemic-epidemic modelling in R

• Applicant: Dr. Sebastian Meyer
• Subject Area: Epidemiology and Medical Biometry/Statistics
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 528691398

Infectious diseases are challenging public health, the COVID-19 pandemic being just one recent example. Health authorities and epidemiologists are faced with questions on vaccine efficacy, whether there are particularly susceptible subgroups of the population, whether some serotypes are more infectious than others, and how fast and broad the disease is spreading. Epidemic modelling is central to our understanding of infectious disease spread, for example, of seasonal or environmental effects and reproduction numbers. Statistical approaches naturally address uncertainty both in parameters of interest and of incidence forecasts. With a regression-like analysis tool in mind, so-called endemic-epidemic models have been developed and implemented in the R package "surveillance" during 2011-2017: a multivariate time-series model for count data ("HHH"), and a spatio-temporal point process model for individual-level data ("twinstim"). Both implementations have proven useful in their original applications, in many epidemiological studies conducted by other research groups, including for COVID-19, and even for criminological models. However, several recent extensions of these implementations are not yet integrated for re-use by third parties. For example, HHH lacks support for a serial interval distribution and zero-inflated counts, both are only implemented in partial forks of "surveillance" that are opaque to non-developers and not available from CRAN. For "twinstim", the Rayleigh kernel for the serial interval distribution and pre-specified background rates are not readily available. Both models still fail to support the new spatial data classes from the "sf" package. Recognizing software as a first-class citizen in research, this project aims to revamp and split the existing implementations into two well-documented, broadly tested, open-source R packages, fully dedicated to these two statistical modelling frameworks for infectious disease spread and other epidemic phenomena.

• DFG Programme: Research Grants