A Negative Binomial Process Model of the 2020–2021 COVID-19 Epidemic in Rhode Island

• Activity Number: 84 - Advances in Spatio-Temporal Statistics with Applications to Environmental Data
• Type: Topic-Contributed
• Date/Time: Monday, August 9, 2021 : 10:00 AM to 11:50 AM
• Sponsor: Section on Statistics and the Environment
• Author(s): Nathan Wikle* and Ephraim Hanks and Maciej Boni
• Companies: Pennsylvania State University and Penn State University and Penn State University
• Keywords: mechanistic modeling; COVID-19; negative binomial process

Abstract

On 01 March 2020, the Rhode Island Department of Health (RIDOH) announced the first positive cases of coronavirus disease 2019 (COVID-19), signaling the beginning of the COVID-19 epidemic in Rhode Island. Over the course of the epidemic, the RIDOH has released data on COVID-19 infection and clinical progression. Using these heterogeneous temporal data sources, we develop a mechanistic, compartmental model of the COVID-19 epidemic, where transmission behavior and clinical progression change with time and across age groups. In particular, we propose a time-inhomogeneous negative binomial process model for emerging epidemics, in which the process intensity is determined by the mechanistic transmission model. This formulation provides a compelling joint likelihood model for eleven data streams, accounting for missing and incomplete data, dependent data, and sources of overdispersion. The resulting model fit is competitive with published mechanistic and phenomenological models. We also discuss possible extensions to a spatio-temporal infectious disease model, where similar data quality challenges are encountered.