Pool regression fits across posterior draws of correct-match classifications

Use posterior draws of the latent match indicators from survregMixBayes() to repeatedly identify which records are treated as correct matches, refit a Cox proportional hazards model on those records, and pool the resulting estimates using multiple-imputation pooling rules.

Each retained posterior draw defines one subset of records classified as correct matches. The function fits the specified survival::coxph() model to that subset, extracts the estimated coefficients and covariance matrix, and combines the results across draws using Rubin's rules.

Usage

# S3 method for class 'survMixBayes'
mi_with(
  object,
  data,
  formula,
  min_n = NULL,
  quietly = TRUE,
  ties = "efron",
  ...
)

Arguments

object

A survMixBayes model object containing posterior draws of the latent match indicators.

data

A data.frame with all candidate records in the same row order as used in the model.

formula

Model formula for refitting on each draw (required), typically of the form survival::Surv(time, event) ~ ....

min_n

Minimum number of records required to fit the model for a given posterior draw. The default is p + 2, where p is the number of non-intercept columns in the model matrix.

quietly

If TRUE, draws that lead to fitting errors are skipped without printing the full error message.

ties

Method for handling tied event times in survival::coxph(). Default is "efron".

...

Additional arguments passed to survival::coxph().

Value

An object of class c("mi_link_pool_survreg", "mi_link_pool") containing pooled coefficient estimates, standard errors, confidence intervals, and related summary information.

Examples

# \donttest{
set.seed(301)
n <- 150
trt <- rbinom(n, 1, 0.5)

# Simulate Weibull AFT data
true_time <- rweibull(n, shape = 1.5, scale = exp(1 + 0.8 * trt))
cens_time <- rexp(n, rate = 0.1)
true_obs_time <- pmin(true_time, cens_time)
true_status <- as.integer(true_time <= cens_time)

# Induce linkage mismatch errors in approximately 20% of records
is_mismatch <- rbinom(n, 1, 0.2)
obs_time <- true_obs_time
obs_status <- true_status
mismatch_idx <- which(is_mismatch == 1)

shuffled <- sample(mismatch_idx)
obs_time[mismatch_idx] <- obs_time[shuffled]
obs_status[mismatch_idx] <- obs_status[shuffled]

linked_df <- data.frame(time = obs_time, status = obs_status, trt = trt)
adj <- adjMixBayes(linked.data = linked_df)

fit <- plsurvreg(
  survival::Surv(time, status) ~ trt,
  dist = "weibull",
  adjustment = adj,
  control = list(iterations = 200, burnin.iterations = 100, seed = 123)
)
#> 
#> SAMPLING FOR MODEL 'survMixBayes_weibull' NOW (CHAIN 1).
#> Chain 1: 
#> Chain 1: Gradient evaluation took 9.4e-05 seconds
#> Chain 1: 1000 transitions using 10 leapfrog steps per transition would take 0.94 seconds.
#> Chain 1: Adjust your expectations accordingly!
#> Chain 1: 
#> Chain 1: 
#> Chain 1: WARNING: There aren't enough warmup iterations to fit the
#> Chain 1:          three stages of adaptation as currently configured.
#> Chain 1:          Reducing each adaptation stage to 15%/75%/10% of
#> Chain 1:          the given number of warmup iterations:
#> Chain 1:            init_buffer = 15
#> Chain 1:            adapt_window = 75
#> Chain 1:            term_buffer = 10
#> Chain 1: 
#> Chain 1: Iteration:   1 / 200 [  0%]  (Warmup)
#> Chain 1: Iteration:  20 / 200 [ 10%]  (Warmup)
#> Chain 1: Iteration:  40 / 200 [ 20%]  (Warmup)
#> Chain 1: Iteration:  60 / 200 [ 30%]  (Warmup)
#> Chain 1: Iteration:  80 / 200 [ 40%]  (Warmup)
#> Chain 1: Iteration: 100 / 200 [ 50%]  (Warmup)
#> Chain 1: Iteration: 101 / 200 [ 50%]  (Sampling)
#> Chain 1: Iteration: 120 / 200 [ 60%]  (Sampling)
#> Chain 1: Iteration: 140 / 200 [ 70%]  (Sampling)
#> Chain 1: Iteration: 160 / 200 [ 80%]  (Sampling)
#> Chain 1: Iteration: 180 / 200 [ 90%]  (Sampling)
#> Chain 1: Iteration: 200 / 200 [100%]  (Sampling)
#> Chain 1: 
#> Chain 1:  Elapsed Time: 0.62 seconds (Warm-up)
#> Chain 1:                0.489 seconds (Sampling)
#> Chain 1:                1.109 seconds (Total)
#> Chain 1: 
#> Warning: The largest R-hat is 1.18, indicating chains have not mixed.
#> Running the chains for more iterations may help. See
#> https://mc-stan.org/misc/warnings.html#r-hat
#> Warning: Bulk Effective Samples Size (ESS) is too low, indicating posterior means and medians may be unreliable.
#> Running the chains for more iterations may help. See
#> https://mc-stan.org/misc/warnings.html#bulk-ess
#> Warning: Tail Effective Samples Size (ESS) is too low, indicating posterior variances and tail quantiles may be unreliable.
#> Running the chains for more iterations may help. See
#> https://mc-stan.org/misc/warnings.html#tail-ess
#> 
#>     ......................................................................................
#>     . Method                         Time (sec)           Status                         . 
#>     ......................................................................................
#>     . ECR-ITERATIVE-1                0.096                Converged (2 iterations)       . 
#>     ......................................................................................
#> 
#>     Relabelling all methods according to method ECR-ITERATIVE-1 ... done!
#>     Retrieve the 1 permutation arrays by typing:
#>         [...]$permutations$"ECR-ITERATIVE-1"
#>     Retrieve the 1 best clusterings: [...]$clusters
#>     Retrieve the 1 CPU times: [...]$timings
#>     Retrieve the 1 X 1 similarity matrix: [...]$similarity
#>     Label switching finished. Total time: 0.1 seconds. 

pooled_obj <- mi_with(
  object = fit,
  data = linked_df,
  formula = survival::Surv(time, status) ~ trt
)

print(pooled_obj)
#> Pooled Cox regression results across posterior match classifications:
#>   Retained imputations (m): 100 
#>   Mixture model distribution: weibull 
#>   Refit model: coxph
#> 
#>     Estimate Std.Error   CI.lwr  CI.upr       df
#> trt -0.97202   0.40304 -1.76665 -0.1774 205.7726
# }