Computational reproducibility is an important part of any computational science as the paper is typically a summarised version of the experiments with the traditional methods section insufficient for a reader to reproduce the results. Often, critical details relating to data pre-processing through to simulation are described imprecisely, and can only be uncovered by analysing code.
Objective:
To assess the computational reproducibility of COVID-19 infectious disease modeling articles due to their potential significance and translational impact.
Findings:
- 88 out of 100 randomly sampled studies Infectious disease models were not computationally reproducible.
- 67 out of 100 top cited Infectious disease models were not computationally reproducible.
- Journals mandating data release are significantly associated with code release.
First published: Feb 10, 2024
The Australian Institute of Tropical Health and Medicine (AITHM)
The [Infectious Disease Modelling and Epidemiology Group](https://www.aithm.jcu.edu.au/research/research-projects-and-groups/covid-19/), at the [ Australian Institute of Tropical Health and Medicine (AITHM)](https://www.aithm.jcu.edu.au/), is dedicated to examining and quantifying the dynamics of infectious diseases of interest to our community in Tropical Northern Australia, the Ind-Pacific Region, and the world. We build models aimed at improved understanding of disease processes and public health policy decision support.
