FG_AugmentCifstrata: Augmentation for Fine-Gray model based on stratified NPMLE...

In mets: Analysis of Multivariate Event Times

Augmentation for Fine-Gray model based on stratified NPMLE Cif (Aalen-Johansen)

Description

Computes the augmentation term for each individual as well as the sum

A(β) = \int H(t,X,β) \frac{F_2^*(t,s)}{S^*(t,s)} \frac{1}{G_c(t)} dM_c

with

H(t,X,β) = \int_t^∞ (X - E(β,t) ) G_c(t) dΛ_1^*i(t,s)

using a KM for

G_c(t)

and a working model for cumulative baseline related to

F_1^*(t,s)

and

s

is strata,

S^*(t,s) = 1 - F_1^*(t,s) - F_2^*(t,s)

, and

E(β^p,t)

is given. Assumes that no strata for baseline of ine-Gay model that is augmented.

Usage

FG_AugmentCifstrata(
  formula,
  data = data,
  E = NULL,
  cause = NULL,
  cens.code = 0,
  km = TRUE,
  case.weights = NULL,
  weights = NULL,
  offset = NULL,
  ...
)

Arguments

formula	formula with 'Event', strata model for CIF given by strata, and strataC specifies censoring strata
data	data frame
E	from FG-model
cause	of interest
cens.code	code of censoring
km	to use Kaplan-Meier
case.weights	weights for FG score equations (that follow dN_1)
weights	weights for FG score equations
offset	offsets for FG model
...	Additional arguments to lower level funtions

Details

After a couple of iterations we end up with a solution of

\int (X - E(β) ) Y_1(t) w(t) dM_1 + A(β)

the augmented FG-score.

Standard errors computed under assumption of correct

G_c

model.

Author(s)

Thomas Scheike

Examples

set.seed(100)
rho1 <- 0.2; rho2 <- 10
n <- 400
beta=c(0.0,-0.1,-0.5,0.3)
dats <- simul.cifs(n,rho1,rho2,beta,rc=0.2)
dtable(dats,~status)
dsort(dats) <- ~time
fg <- cifreg(Event(time,status)~Z1+Z2,data=dats,cause=1,propodds=NULL)
summary(fg)

fgaugS <- FG_AugmentCifstrata(Event(time,status)~Z1+Z2+strata(Z1,Z2),data=dats,cause=1,E=fg$E)
summary(fgaugS)
fgaugS2 <- FG_AugmentCifstrata(Event(time,status)~Z1+Z2+strata(Z1,Z2),data=dats,cause=1,E=fgaugS$E)
summary(fgaugS2)

mets documentation built on Jan. 17, 2023, 5:12 p.m.