This repository contains a simple model used for forecasting hospitalizations in California counties in January 2022.
- Hospitalization and case data come from the California Open Data Portal.
- Delays for case reporting are estimated using line lists provided by the Orange County Health Care Agency.
- Prevalence of variants in California is estimated from from GISAID data via the Bedford Lab
We use est_omicron_cases = reported_cases / proportion_cases_expected * proportion_omicron and est_other_cases = reported_cases / proportion_cases_expected * (1 - proportion_omicron) as data in the model.
The model is a S-E-I-R type model startified by omicron vs non-omicrons with hospitalization compartments. The model is fit with fit_model.jl using Turing.jl.
Priors and posteriors for the model parameters are presented in prior_post_plots.pdf.
Results for each county are presented in post_pred_plots.pdf.