What question did this study set out to answer?

To evaluate the predictive skill of individual and multimodel ensemble probabilistic forecasts of COVID-19 mortality at state and national levels in the United States.

April 8, 2022Proceedings of the National Academy of SciencesOpen Access

Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the United States

View Full Paper

Ask AI

Bookmark

To evaluate the predictive skill of individual and multimodel ensemble probabilistic forecasts of COVID-19 mortality at state and national levels in the United States.
Synthesized tens of millions of predictions submitted to the US COVID-19 Forecast Hub from over 90 academic, industry, and independent research groups between April 2020 and October 2021.
Evaluated probabilistic forecasts of incident COVID-19 deaths across multiple horizons (up to 20 weeks), comparing 27 consistently reporting individual models and a weekly multimodel ensemble against a naïve baseline model.
The multimodel ensemble forecast consistently achieved the highest probabilistic accuracy for incident COVID-19 deaths at both state and national levels across the evaluation period.
Two-thirds of the 27 evaluated individual models outperformed the naïve baseline model, though individual accuracy varied substantially across time, geographic regions, and forecast horizons.
Forecast performance degraded with longer horizons, resulting in probabilistic error at a 20-week horizon that was three to five times higher than at a 1-week horizon.

Discussion

No takes yet. Share an insight, caveat, or question.

Implication

Modeling evaluation reveals multimodel ensembles consistently provide superior COVID-19 mortality predictions in the United States, highlighting the value of collaborative outbreak forecasting.

Cite This Study

Cramer et al. (2022) studied this question.

synapsesocial.com/papers/6aaf6505559bae6622110f7d https://doi.org/10.1073/pnas.2113561119

View Full Paper

Ask AI

Bookmark

Implication

Modeling evaluation reveals multimodel ensembles consistently provide superior COVID-19 mortality predictions in the United States, highlighting the value of collaborative outbreak forecasting.

Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the United States

Key Points

Discussion

Implication

Cite This Study

Evaluation of individual and ensemble probabilistic forecasts of COVID-19 mortality in the United States

Key Points

Discussion

Implication

Cite This Study

Also Consider

Also Consider