EpiFlowML: A modular framework for standardized and reproducible epidemiological forecasting
SoftwareX · 2026-09 · 35 · 102952
DOI: 10.1016/j.softx.2026.102952 ↗Kurz gefasst
Fachartikel in der Zeitschrift SoftwareX, 2026-09. Verfasst von Amanda J. Meadows und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 19 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Bill and Melinda Gates Foundation.
Abstract (Originalsprache)
EpiFlowML is a modular forecasting framework designed to standardize how epidemiological time series are prepared, modeled, and evaluated across multiple geographies and data sources. The system provides a harmonized workflow for transforming heterogeneous datasets into consistent, analysis-ready formats, fitting and comparing a range of machine learning (ML) models, and producing structured, reproducible outputs. Each module can be configured independently, allowing users to adapt the workflow to different pathogens or health indicators without modifying core code. The framework integrates standardized data ingestion, preprocessing, model fitting, and evaluation routines that generate uniform outputs suitable for comparison and ensemble development. By aligning these components through configuration files and metadata tracking, EpiFlowML supports transparent, reproducible forecasting across diverse epidemiological contexts. The increasing complexity and volume of forecasting models underscores the need for such standardization (1, 2). A global measles forecasting task was used to demonstrate the framework's functionality. This use case highlights the value of harmonized data structures and standardized evaluation in improving reproducibility and interpretability, as we were able to model 111 countries in the same workflow. The framework and supporting code are fully open source under the Creative Commons 4.0 License, enabling adaptation to other diseases or forecasting applications.
Förderung
- Bill and Melinda Gates Foundation
Zitieren
APA
Meadows, A. J., Schlitt, J., Sridharan, V., Stephenson, N., & Madhav, N. K. (2026). EpiFlowML: A modular framework for standardized and reproducible epidemiological forecasting. SoftwareX, 35, 102952. https://doi.org/10.1016/j.softx.2026.102952
Vancouver
Meadows AJ, Schlitt J, Sridharan V, Stephenson N, Madhav NK. EpiFlowML: A modular framework for standardized and reproducible epidemiological forecasting. SoftwareX. 2026;35:102952. doi:10.1016/j.softx.2026.102952
BibTeX
@article{Meadows2026,
author = {Meadows, Amanda J. and Schlitt, James and Sridharan, Vikram and Stephenson, Nicole and Madhav, Nita K.},
title = {EpiFlowML: A modular framework for standardized and reproducible epidemiological forecasting},
journal = {SoftwareX},
year = {2026},
volume = {35},
pages = {102952},
doi = {10.1016/j.softx.2026.102952}
}
EUVIMED