Stan fitting and draw extraction now use flexstanr. Configure automatic, scheduler-aware within-chain threading with stan_options(threading = TRUE).
First release.
Fit Bayesian compartmental infection models from individual-level outcome data, using Stan via ‘rstan’. Models are composed from an infection process and an observation process and fit with a single call.
Infection-process building blocks: transmit() and progress() define transitions, assembled with make_infection_model(). Susceptible-infected- recovered (SIR), split symptomatic/asymptomatic (SIIR), and similar multi-compartment structures are supported.
make_observation_model() defines the observation process (per-compartment detection probabilities for one or more tests).
run_model() fits the assembled model to the data and returns posterior draws on the natural (model) scale.
Bundled simulated datasets and fitted-draw fixtures for examples and the vignettes: sir, sir_cov, siir, and the sir_res, sir_cov_res, siir_res posterior draws.
Vignettes: “SIR” and “multiple infection compartments”.