AIM (version 1.01)

cv.cox.interaction: Cross-validation in the interaction Cox AIM

Description

Cross-validation for selecting the number of binary rules in interaction AIM with survival outcomes in the context of Cox regression.

Usage

cv.cox.interaction(x, trt, y, status, K.cv=5, num.replicate=1, nsteps, mincut=0.1, backfit=F, maxnumcut=1, dirp=0)

Arguments

x

n by p matrix. The covariate matrix

trt

n vector. The treatment indicator

y

n vector. The observed follow-up time

status

n 0/1 vector. The status indicator. 1=failure and 0=alive.

K.cv

K.cv-fold cross validation

num.replicate

number of independent replications of K-fold cross validations.

nsteps

the maximum number of binary rules to be included in the index

backfit

T/F. Whether the existing split points are adjusted after including a new binary rule

mincut

the minimum cutting proportion for the binary rule at either end. It typically is between 0 and 0.2.

maxnumcut

the maximum number of binary splits per predictor

dirp

p vector. The given direction of the binary split for each of the p predictors. 0 represents "no pre-given direction"; 1 represents "(x>cut)"; -1 represents "(x

Value

cv.cox.interaction returns

kmax

the optimal number of binary rules based the cross-validation

meanscore

nsteps-vector. The cross-validated partial likelihood score test statistics (significant at 0.05, if greater than 1.96) for the treatment*index interaction.

pvfit.score

nsteps-vector. The pre-validated partial likelihood score test statistics (significant at 0.05, if greater than 1.96) for the treatment*index interaction.

preval

nsteps by n matrix. Pre-validated fits for individual observation

References

L Tian and R Tibshirani Adaptive index models for marker-based risk stratification, Tech Report, available at http://www-stat.stanford.edu/~tibs/AIM. R Tibshirani and B Efron, Pre-validation and inference in microarrays, Statist. Appl. Genet. Mol. Biol., 1:1-18, 2002.

Details

cv.cox.interaction implements K-fold cross-validation for the interaction Cox AIM. It estimates the partial likelihood score test statistics for testing the treatment*index interaction in the test set. It also provides pre-validated fits for each observation and pre-validated partial likelihood score test statistics. The output can be used to select the optimal number of binary rules.

Examples

## generate data
set.seed(1)

n=400
p=10
x=matrix(rnorm(n*p), n, p)
z=(x[,1]<0.2)+(x[,5]>0.2)
beta=1
trt=rbinom(n,1,0.5)
fail.time=rexp(n)*exp(-beta*z*trt)
cen.time=rexp(n)*1.25
y=pmin(fail.time, cen.time)
y=round(y*10)/10
delta=1*(fail.time<cen.time)



## cross-validate the interaction Cox AIM model 
a=cv.cox.interaction(x, trt, y, delta, nsteps=10, K.cv=4, num.replicate=5)
 
## examine the score test statistics for the interaction in the test set 
par(mfrow=c(1,2))
plot(a$meanscore, type="l")
plot(a$pvfit.score, type="l")

## construct the index with the optimal number of binary rules 
k.opt=a$kmax
a=cox.interaction(x, trt, y, delta, nsteps=k.opt)
print(a)