EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

The influence of ensemble size and composition on the performance of combined real-time COVID-19 forecasts

Friederike Becker iD, Katharine Sherratt, Nikos I Bosse, Sebastian Funk

PLOS Computational Biology · 03.11.2026 · 22 (11) · e1014908

DOI: 10.1371/journal.pcbi.1014908 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS Computational Biology, online erschienen am 03.11.2026. Verfasst von Friederike Becker und 3 weiteren Autor:innen. Beteiligte Einrichtung: Institute of Statistics, Karlsruhe Institute of Technology, Karlsruhe, Germany. Hauptthemen laut MEDLINE: COVID-19, Coronavirus Infections. Die Arbeit nennt 26 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Helmholtz Association, Wellcome Trust, National Institute for Health Research Health Protection Research Unit.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

During infectious disease outbreaks, short-term forecasts can play an important role for both decision makers and the general public. While previous research has shown that combining individual forecasts into an ensemble improves accuracy and consistency, practical guidance for organisers of multi-model prediction platforms on how to construct an ensemble has been scarce. In particular, it is not entirely clear how ensemble performance relates to the size of the underlying model pool, a relevant question when relying on voluntary contributions from modelling teams that face competing priorities. Furthermore, the exact composition of an ensemble forecast may influence its performance. Ensembles can either include all models equally or, alternatively, discriminate based on past performance or other characteristics. Using data from the European COVID-19 Forecast Hub we investigated these questions, with the aim of offering practical guidance to organisers of multi-model prediction platforms during infectious disease outbreaks. We found that including more models both improved and stabilised aggregate ensemble performance, while selecting for better component models did not yield any particular advantage. Diversity among models, whether measured numerically or qualitatively, did not have a clear impact on ensemble performance. These results suggest that for those soliciting contributions to collaborative ensembles there are more obvious gains to be made from increasing participation to moderate levels than from optimising component models.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

COVID-19Coronavirus InfectionsHumansEnsemble LearningForecastingPandemicsSARS-CoV-2Computational BiologyDisease OutbreaksPrediction AlgorithmsEurope

Autor:innen und Institutionen

  • Institute of Statistics, Karlsruhe Institute of Technology, Karlsruhe, Germany.
  • Centre for Mathematical Modelling of Infectious Diseases, London School of Hygiene & Tropical Medicine, London, United Kingdom.
  • Department of Infectious Disease Epidemiology and Dynamics, London School of Hygiene & Tropical Medicine, London, United Kingdom.

ORCID-Profile: Friederike Becker

Förderung

  • Helmholtz Association
  • Wellcome Trust
  • National Institute for Health Research Health Protection Research Unit

Zitieren

APA

Becker, F., Sherratt, K., Bosse, N. I., & Funk, S. (2026). The influence of ensemble size and composition on the performance of combined real-time COVID-19 forecasts. PLOS Computational Biology, 22(11), e1014908. https://doi.org/10.1371/journal.pcbi.1014908

Vancouver

Becker F, Sherratt K, Bosse NI, Funk S. The influence of ensemble size and composition on the performance of combined real-time COVID-19 forecasts. PLOS Computational Biology. 2026;22(11):e1014908. doi:10.1371/journal.pcbi.1014908

BibTeX
@article{Becker2026,
  author = {Becker, Friederike and Sherratt, Katharine and Bosse, Nikos I and Funk, Sebastian},
  title = {The influence of ensemble size and composition on the performance of combined real-time COVID-19 forecasts},
  journal = {PLOS Computational Biology},
  year = {2026},
  volume = {22},
  number = {11},
  pages = {e1014908},
  doi = {10.1371/journal.pcbi.1014908}
}
Kostenlos starten →