Correcting delayed reporting of COVID-19 using the generalized-Dirichlet-multinomial method

Stoner, O. , Halliday, A. and Economou, T. (2023) Correcting delayed reporting of COVID-19 using the generalized-Dirichlet-multinomial method. Biometrics, (doi: 10.1111/biom.13810) (PMID:36484382) (PMCID:PMC9877609) (Early Online Publication)

Abstract

The COVID-19 pandemic has highlighted delayed reporting as a significant impediment to effective disease surveillance and decision-making. In the absence of timely data, statistical models which account for delays can be adopted to nowcast and forecast cases or deaths. We discuss the four key sources of systematic and random variability in available data for COVID-19 and other diseases, and critically evaluate current state-of-the-art methods with respect to appropriately separating and capturing this variability. We propose a general hierarchical approach to correcting delayed reporting of COVID-19 and apply this to daily English hospital deaths, resulting in a flexible prediction tool which could be used to better inform pandemic decision-making. We compare this approach to competing models with respect to theoretical flexibility and quantitative metrics from a 15-month rolling prediction experiment imitating a realistic operational scenario. Based on consistent leads in predictive accuracy, bias, and precision, we argue that this approach is an attractive option for correcting delayed reporting of COVID-19 and future epidemics.

• Item Type: Articles
• Status: Early Online Publication
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Stoner, Dr Oliver and Halliday, Alba
• Authors: Stoner, O., Halliday, A., and Economou, T.
• College/School: College of Science and Engineering > School of Mathematics and Statistics > Statistics
• Journal Name: Biometrics
• Publisher: Wiley
• ISSN: 0006-341X
• ISSN (Online): 1541-0420
• Published Online: 09 December 2022
• Copyright Holders: Copyright © 2022 The Author
• First Published: First published in Biometrics 2023
• Publisher Policy: Reproduced under a Creative Commons licence