Fit Multilevel Unanchored Meta-Regression Model

Fit an ML-UMR model to perform population-adjusted indirect comparison between treatments in disconnected networks using IPD and AgD.

Usage

mlumr(
  data,
  spfa = TRUE,
  prior_intercept = prior_normal(0, 10),
  prior_beta = prior_normal(0, 10),
  prior_sigma = prior_normal(0, 10),
  iter_warmup = 1000,
  iter_sampling = 2000,
  chains = 4,
  parallel_chains = chains,
  refresh = 100,
  seed = NULL,
  adapt_delta = 0.95,
  max_treedepth = 15,
  ...
)

Arguments

data: An unanchored_data object with integration points
spfa: Logical. If TRUE, invoke Shared Prognostic Factor Assumption (regression coefficients shared between treatments). If FALSE, allow treatment-specific coefficients (effect modification). Default TRUE.
prior_intercept: Prior for treatment intercepts (see prior_normal)
prior_beta: Prior for regression coefficients
prior_sigma: Prior for residual standard deviation (normal likelihood only, default prior_normal(0, 10)). Ignored for binomial and Poisson likelihoods.
iter_warmup: Number of warmup iterations (default 1000)
iter_sampling: Number of sampling iterations (default 2000)
chains: Number of MCMC chains (default 4)
parallel_chains: Number of chains to run in parallel (default: all)
refresh: How often to print progress (0 = no output, default 100)
seed: Random seed for reproducibility
adapt_delta: Target acceptance rate (default 0.95)
max_treedepth: Maximum tree depth (default 15)
...: Additional arguments passed to cmdstanr::sample()

Value

An object of class mlumr_fit

Examples

if (FALSE) { # \dontrun{
# Fit model with SPFA
fit_spfa <- mlumr(
  data = network,
  spfa = TRUE,
  iter_warmup = 1000,
  iter_sampling = 2000,
  chains = 4
)

# Fit model with relaxed SPFA
fit_relaxed <- mlumr(
  data = network,
  spfa = FALSE
)
} # }