survMTP.estimate: Accuracy measures for a risk prediction marker under...

In mdbrown/survMarkerTwoPhase: Evaluate the prognostic accuracy of a marker under two phase designs

Description

This function estimates the AUC, TPR(c), FPR(c), PPV(c), and NPV(c) for for a specific timepoint and marker cutoff value c with survival data from a case-cohort subcohort design.

Usage

  survMTP.cch(time, event, marker, 
                             weights, 
                             subcoh,
                             data, 
                             estimation.method = 'NP', 
                             predict.time, 
                             marker.cutpoint = 'median')

Arguments

time	time to event
event	indicator for the status of event of interest. event = 0 for censored observations, and event = 1 for event of interest.
marker	marker variable of interest
weights	sample weights from a case-cohort sample design. Should be equal to 1/Pr(sampled from cohort).
subcoh	indicator for subjects included in the subcohort sample (1=included, 0=not included)
data	data frame in which to look for input variables. This should be the full cohort data, including time, event, weights, subcoh, and marker values when subcoh = 1.
estimation.method	should non-parametric ('NP') or semi-parametric ('SP') estimates be calculated. Semi-parametric methods assume a proportional hazards model.
predict.time	numeric value of the timepoint of interest for which to estimate the risk measures
marker.cutpoint	numeric value indicating the value of the cutpoint 'c' at which to estimate 'FPR', 'TPR', 'NPV' or 'PPV'. default is 'median' which takes cutpoint as the marker median.

Value

a list with components

estimates	point estimates for risk measures
model.fit	*only returned if estimation.method = 'SP'* object of type 'coxph' from fitting the model coxph(Surv(time, event)~Y)
marker.cutoff, estimation.method, predict.time	function inputs

References

1. Liu D, Cai T, Zheng Y. Evaluating the predictive value of biomarkers with stratified case-cohort design. *Biometrics* 2012, 4: 1219-1227.

2. Pepe MS, Zheng Y, Jin Y. Evaluating the ROC performance of markers for future events. *Lifetime Data Analysis.* 2008, 14: 86-113.

3. Zheng Y, Cai T, Pepe MS, Levy, W. Time-dependent predictive values of prognostic biomarkers with failure time outcome. *JASA* 2008, 103: 362-368.

4. Cai T. and Zheng Y . Resampling Procedures for Making Inference under Nested Case-control Studies. *JASA* 2013 (in press).

5. Cai T and Zheng Y/*, Evaluating Prognostic Accuracy of Biomarkers under nested case-control studies. *Biostatistics* 2012, 13,1, 89-100.

(* equal contributor and corresponding author).

Examples

#simulated data for illustration
data(SimData)

#generate a sub-cohort from SimData
set.seed(12321)
#create a sample index. 1 if sampled, 0 if not
N <- nrow(SimData)
sampleInd <- rep(0, N)

# sample all with observed failure time. (200 individuals)
sampleInd[SimData$status==1] <- 1

#sample 150 more observations from the entire data set without replacement
sampleInd[sample(1:N, 150)] <- 1

table(sampleInd) #total number of subcohort is 293 

cohortData_cch <- SimData

#first calculate the Pr(Sampled from cohort) for each observation
sampleProb <- numeric(500)

#all non-censored observations were sampled, so their sample probability is 1
sampleProb[cohortData_cch$status==1] <- 1 
#all other individuals had a 150/N chance to be sampled 
sampleProb[cohortData_cch$status==0] <- 150/N

#the sample weights are 1/(probability of being sampled)
cohortData_cch$weights <- 1/sampleProb

#indicator of inclusion in the subcohort
cohortData_cch$subcohort = sampleInd   

#estimate accuracy measures using non-parametric estimates
#by setting estimation.method = "NP"
survMTP.cch(time =survTime, 
            event = status, 
            marker = Y,
            weights = weights,
            subcoh = subcohort,
            data = cohortData_cch, 
            estimation.method = "NP",  
            predict.time = 2, 
            marker.cutpoint = 0)

#estimate accuracy measures using semi-parametric estimates
survMTP.cch(time =survTime, 
            event = status, 
            marker = Y,
            weights = weights,
            subcoh = subcohort,
            data = cohortData_cch, 
            estimation.method = "SP",  
            predict.time = 2, 
            marker.cutpoint = 0)

mdbrown/survMarkerTwoPhase documentation built on May 22, 2019, 3:23 p.m.