The source and corrected draft and a small case study explaining how to code Andrew’s and my model in Stan is available in our diagnostic-testing repo on GitHub.
Every statistical model is an error model in some sense. Specifically, our model lets you perform inference using a hierarchical meta-analysis of the sensitivity and specificity at a specific existing site or at a newly created test site.
Code for published paper
that is to some (unclear) extent being used by the model relied upon by Dallas County health officials
Has anyone tried Fractional differential equations on STAN for modeling diseases?
The EpiNow2 R package uses Stan. It’s being used for national and subnational R/growth rate estimates which themselves have been used in a number of publications, e.g. Li et al. (2020), Davies et al. (2021), and others.
Stan has also been used for nowcasts and projections of COVID-19 in the UK , to estimate epidemiological parameters from repeated cross-sectional prevalence studies, to perform nowcasting of censored case counts, and to estimate transmission advantage and generation intervals of new variants.
Sharing my 2 main COVID-19 papers, using STAN.
To my knowledge the best estimates of age-stratified COVID-19 severity, estimated through a literature meta-analysis. STAN was instrumental to put together information sources with widely different uncertainties https://link.springer.com/article/10.1186/s12879-022-07262-0 Open code and data here: GitHub - dherrera1911/estimate_covid_severity
A meta-analysis of time-varying sensitivity in COVID-19 serological assays that can serve as a guide to correct for this source of bias in the literature. https://www.medrxiv.org/content/10.1101/2022.09.08.22279731v1 Code and data here: GitHub - dherrera1911/seroreversion_metaanalysis
Thanks!
Here’s our preprint, currently under review:
Bayesian analysis of diffusion-driven multi-type epidemic models with application to COVID-19
The Bernadette R package [CRAN link] accompanies the work.
Lampros