Extract Survival Outcome Components and Predictors — extract_surv

Helper utility that molds a survival formula into clean components, through the same hardhat formula blueprint used by coxnet (one-hot dummy encoding, no intercept), supporting both 2-parameter Surv(time, status) and 3-parameter start/stop Surv(tstart, tstop, status) counting process structures.

Usage

extract_surv_components(data, outcome, subject_id = NULL)

Arguments

data: A data frame.
outcome: A survival formula or Surv expression.
subject_id: Optional subject identifier column name; dropped from `data` before molding, so it is never treated as a predictor.

Value

A list with elements surv_obj, time, start, status, type, predictors (a numeric tibble, one-hot encoded), pred_names, raw_predictors (the original predictor columns, before encoding, which is what the models' predict() methods and the explainers take), formula, data (the molded, `subject_id`-free data frame), and blueprint (the hardhat blueprint used, for forge).