Skip to contents

Compute Posterior Resistance Profile Probabilities via MVN Simulation

Source: R/daly_resistance_profiles.R
compute_event_profile_probabilities.Rd

For each posterior draw, constructs the event-level linear predictor \(\mu_e\) using fixed effects and correlated random effects from the fitted model, simulates latent \(Z_e = \mu_e + L_\Omega\,\varepsilon_e\), \(\varepsilon_e \sim N(0,I_D)\), converts to binary resistance profiles, and accumulates profile probabilities. All \(2^D\) profiles appear in the output; profiles not observed in simulation receive probability 0.

Usage

compute_event_profile_probabilities(
  fitted_model,
  n_posterior_draws_for_profiles = 2000L,
  outcome_col = NULL,
  nonfatal_values = c("Discharged", "Survived", "Alive", "discharged", "survived",
    "alive"),
  seed = 123L
)

Arguments

fitted_model

List returned by fit_bayesian_multivariate_probit().

n_posterior_draws_for_profiles

Integer. Number of posterior draws to use for profile simulation. Subsampled without replacement when total draws exceed this value. Each draw generates one simulated latent profile per event; rare profiles may be underestimated when this is small. For finer Monte Carlo resolution of rare profiles, consider increasing this value or adding n_predictive_replicates_per_draw in a future extension. Default 2000L.

outcome_col

Character or NULL. Patient outcome column in fitted_model$event_metadata. When NULL, all events are treated as having a known outcome and R_NF is not computed separately.

nonfatal_values

Character vector. Outcome values for the non-fatal cohort (R_NF). Default covers common discharge/survival labels.

seed

Integer. Random seed for MVN simulation. Default 123L.

Value

Named list: event_profiles (event-level posterior means) and aggregate_draws (per-draw R_ALL and R_NF per hospital x pathogen x profile, used by aggregate_profiles_for_daly() for credible intervals). Both tibbles contain all \(2^D\) profiles per hospital-pathogen pair.

Details

Estimand: The posterior predictive distribution over the observed event case-mix – the profile distribution you would see if you drew a new event uniformly from the set of observed events (same covariate distribution as the data). This is labelled "observed_stewardship_event_mix".