BuyseTTEM: Time to Event Model

In BuyseTest: Generalized Pairwise Comparisons

Time to Event Model

Description

Pre-compute quantities of a time to event model useful for predictions. Only does something for prodlim objects.

Usage

BuyseTTEM(object, ...)

## S3 method for class 'formula'
BuyseTTEM(object, treatment, iid, iid.surv = "exp", ...)

## S3 method for class 'prodlim'
BuyseTTEM(object, treatment, iid, iid.surv = "exp", ...)

## S3 method for class 'survreg'
BuyseTTEM(object, treatment, n.grid = 1000, iid, ...)

## S3 method for class 'BuyseTTEM'
BuyseTTEM(object, ...)

Arguments

object	time to event model.
...	For compatibility with the generic method.
treatment	[list or character] List containing the value of the treatment variable for each individual or name of the treatment variable. The former is necessary when the survival model do not dependent on treatment.
iid	[logical] Should the iid decomposition of the predictions be output.
iid.surv	[character] Estimator of the survival used when computing the influence function. Can be the product limit estimator ("prodlim") or an exponential approximation ("exp", same as in riskRegression::predictCoxPL).
n.grid	[integer, >0] Number of timepoints used to discretize the time scale. Not relevant for prodlim objects.

Value

An S3 object of class BuyseTTEM.

Examples

library(prodlim)
library(data.table)

tau <- seq(0,3,length.out=10)

#### survival case ####
set.seed(10)
df.data <- simBuyseTest(1e2, n.strata = 2)

e.prodlim <- prodlim(Hist(eventtime,status)~treatment+strata, data = df.data)
## plot(e.prodlim)

e.prodlim2 <- BuyseTTEM(e.prodlim, treatment = "treatment", iid = TRUE)

predict(e.prodlim2, time = tau, treatment = "T", strata = "a")
predict(e.prodlim, times = tau, newdata = data.frame(treatment = "T", strata = "a"))

predict(e.prodlim2, time = tau, treatment = "C", strata = "a")
predict(e.prodlim, times = tau, newdata = data.frame(treatment = "C", strata = "a"))

#### competing risk case ####
df.dataCR <- copy(df.data)
df.dataCR$status <- rbinom(NROW(df.dataCR), prob = 0.5, size = 2)

e.prodlimCR <- prodlim(Hist(eventtime,status)~treatment+strata, data = df.dataCR)
## plot(e.prodlimCR)

e.prodlimCR2 <- BuyseTTEM(e.prodlimCR, treatment = "treatment", iid = TRUE)

predict(e.prodlimCR2, time = tau, treatment = "T", strata = "a")
predict(e.prodlimCR, times = tau, newdata = data.frame(treatment = "T", strata = "a"), cause = 1)

predict(e.prodlimCR2, time = tau, treatment = "C", strata = "a")
predict(e.prodlimCR, times = tau, newdata = data.frame(treatment = "C", strata = "a"), cause = 1)

BuyseTest documentation built on April 12, 2025, 1:54 a.m.