> ## Content Index
> Fetch the complete content index at: https://www.coios.me/llms.txt
> Use this file to discover other available public pages before exploring further.

# Ensemble models of expected US mortality were more accurate than single models
- URL: https://www.coios.me/ensemble-models-of-expected-us-mortality-were-more-accurate/
- Published: 2026-10-26T23:31:43.000Z
- Updated: 2026-10-27T07:06:23.000Z
- Description: Using weekly US deaths for 2008–2024, ensemble models gave more accurate expected mortality, with better coverage, than single models.
- Author: Daniel Ryan
- Tags: Data, models and systems, Forecasts and their record, Infection and pandemics, United States, #peer-reviewed, #refines, #item, #source-2026-10, #score-72

The study tested ensemble models for estimating expected mortality, the baseline against which excess deaths are measured. It used weekly death counts for the United States from the National Vital Statistics System, 2008 to 2024\. The authors compared 76 candidate models, with different long-term and seasonal trend specifications, against ensemble models. Ensembles performed more consistently and robustly, with better accuracy and interval coverage than individual models. The abstract does not report the size of these gains in accuracy or coverage.

*Why it is interesting: Excess mortality estimates after COVID-19 have varied with the model and analytic choices used. Pooling many trend specifications reduces how far the expected baseline depends on any one of those choices.*

Source

[Epidemiologic Methods, 24 October 2026](https://doi.org/10.1515/em-2025-0035?ref=coios.me)

Paper

An ensemble modeling approach for estimating expected patterns in mortality in the United States

Authors

Rossen LM, Cisewski JA, Ahmad FB, Anderson RN

DOI

10.1515/em-2025-0035

Type

Peer-reviewed article

Design

Methods comparison of 76 candidate and ensemble models for expected mortality, US weekly NVSS data 2008-2024, evaluated on accuracy and interval coverage

Verdict

Refines prior evidence

Driver

[Forecasts and their record](https://www.coios.me/d-forecasting/)

Driver

[Infection and pandemics](https://www.coios.me/d-infectious-disease/)

Free full text

[Free copy in a repository (submitted manuscript)](https://pmc.ncbi.nlm.nih.gov/articles/PMC13693783/?ref=coios.me)