Summarize Bootstrap Event Counts

Provides summary statistics for event counts across bootstrap iterations, helping assess the reliability of HR estimates when events are sparse.

Usage

summarize_bootstrap_events(boot_results, threshold = 5)

Arguments

boot_results

List. Output from forestsearch_bootstrap_dofuture()

threshold

Integer. Minimum event threshold for flagging low counts (default: 5)

Value

Invisibly returns a list with summary statistics:

threshold

The event threshold used

nb_boots

Total number of bootstrap iterations

n_successful

Number of iterations that found a new subgroup

original_H

List with low event counts for original H on bootstrap samples

original_Hc

List with low event counts for original Hc on bootstrap samples

new_Hstar

List with low event counts for new H* on original data

new_Hcstar

List with low event counts for new Hc* on original data

Details

This function summarizes event counts in four scenarios:

1. ORIGINAL subgroup H evaluated on BOOTSTRAP samples

2. ORIGINAL subgroup Hc evaluated on BOOTSTRAP samples

3. NEW subgroup H* (found in bootstrap) evaluated on ORIGINAL data

4. NEW subgroup Hc* (found in bootstrap) evaluated on ORIGINAL data

Low event counts (below threshold) can lead to unstable HR estimates. This summary helps identify potential issues with sparse events.

Examples

if (FALSE) { # \dontrun{
# After running bootstrap analysis
summarize_bootstrap_events(fs_bc, threshold = 10)
} # }