Skip to contents

Univariable Kaplan-Meier and Cox Screening Engine

Source: R/cbe_km_single.R
km_single.Rd

Pairs a univariable [cox_single()] Cox proportional hazards fit with the matching Kaplan-Meier stratified survival curve for the same predictor, so both are produced from a single call without refitting the Cox model twice.

Usage

km_single(data, outcome = "outcome", feature, conf_level = 0.95)

Arguments

data

A data frame containing survival data and the candidate feature.

outcome

Character string naming the Surv object or outcome column in data (e.g., "outcome" or "Surv(survival_time, status)").

feature

Character string naming the candidate predictor column in data. Factor (or coercible) predictors are used directly as Kaplan-Meier strata; numeric predictors are binned into quartiles for the Kaplan-Meier strata (matching [plot_cox_survival()]'s convention), while the Cox side treats them as continuous.

conf_level

Numeric confidence level (default: 0.95).

Value

An object of class cbe_km containing:

  • cox: The cbe_cox object from [cox_single()].

  • km_fit: The survival::survfit object for the Kaplan-Meier curve.

  • km_tidy: broom::tidy() table of km_fit.

  • direction: "increases" or "decreases", pulled from the Cox hazard ratio direction.

  • summary: One-row-per-stratum tidy summary combining Cox hazard ratios and Kaplan-Meier median survival.

  • feature: Feature name string.

  • var_label: Human-readable variable label.

See also

[cox_single()], [plot_cox_survival()]