Country-report pattern corrections of new cases allow accurate 2-week predictions of COVID-19 evolution with the Gompertz model

Accurate short-term predictions of COVID-19 cases with empirical models allow Health Officials to prepare for hospital contingencies in a two–three week window given the delay between case reporting and the admission of patients in a hospital. We investigate the ability of Gompertz-type empiric mode...

Full description

Saved in:

Bibliographic Details
Published in:	Scientific reports Vol. 14; no. 1; p. 10775
Main Authors:	Villanueva, I., Conesa, D., Català, M., López Cano, C., Perramon-Malavez, A., Molinuevo, D., de Rioja, V. L., López, D., Alonso, S., Cardona, P. J., Montañola-Sales, C., Prats, C., Alvarez-Lacalle, E.
Format:	Journal Article
Language:	English
Published:	London Nature Publishing Group UK 11-05-2024 Nature Publishing Group Nature Portfolio
Subjects:	631/114/2397 692/308/174 Case reports COVID-19 COVID-19 - epidemiology COVID-19 - virology Disease transmission Humanities and Social Sciences Humans Models, Statistical multidisciplinary Predictions SARS-CoV-2 - isolation & purification Science Science (multidisciplinary) Time Factors
Online Access:	Get full text
Tags:	Add Tag No Tags, Be the first to tag this record!

Description
Summary:	Accurate short-term predictions of COVID-19 cases with empirical models allow Health Officials to prepare for hospital contingencies in a two–three week window given the delay between case reporting and the admission of patients in a hospital. We investigate the ability of Gompertz-type empiric models to provide accurate prediction up to two and three weeks to give a large window of preparation in case of a surge in virus transmission. We investigate the stability of the prediction and its accuracy using bi-weekly predictions during the last trimester of 2020 and 2021. Using data from 2020, we show that understanding and correcting for the daily reporting structure of cases in the different countries is key to accomplish accurate predictions. Furthermore, we found that filtering out predictions that are highly unstable to changes in the parameters of the model, which are roughly 20%, reduces strongly the number of predictions that are way-off. The method is then tested for robustness with data from 2021. We found that, for this data, only 1–2% of the one-week predictions were off by more than 50%. This increased to 3% for two-week predictions, and only for three-week predictions it reached 10%.
Bibliography:	ObjectType-Article-1 SourceType-Scholarly Journals-1 ObjectType-Feature-2 content type line 23
ISSN:	2045-2322 2045-2322
DOI:	10.1038/s41598-024-61233-w