Cindex {nftbart}

Calculate the C-index/concordance for survival analysis.

Description

The C-index for survival analysis is the corollary of the c statistic (the area under the Receiver Operating Characteristic curve) for binary outcomes. As a probability, the higher is the C-index, the better is the model discrimination vs. lesser probability values. Similarly, the concordance is calculated like the C-index from z-draws via the posterior predictive distribution restricted to the horizon of the data (a la restricted mean survival time).

Usage

Cindex(risk, times, delta=NULL)

concordance(draws, times, delta=NULL)

Arguments

risk

A vector or prognostic risk scores.

draws

A vector of draws via the posterior predictive distribution restricted to the horizon of the data (a la restricted mean survival time).

times

A vector of failure times.

delta

The corresponding failure time status code: 0, right-censored; 1, failure; or 2, left-censored. Defaults to all failures if not specified.

Value

The return value is the calculated C-index/concordance.

References

Harrell FE, Califf RM, Pryor DB, Lee KL, Rosati RA. (1982) Evaluating the yield of medical tests. JAMA, May 14;247(18):2543-6.

See Also

predict.nft

Examples

data(lung)
N=length(lung$status)

##lung$status: 1=censored, 2=dead
##delta: 0=censored, 1=dead
delta=lung$status-1

## this study reports time in days
times=lung$time
times=times/7  ## weeks

## matrix of covariates
x.train=cbind(lung[ , -(1:3)])
## lung$sex:        Male=1 Female=2

## Not run: 
    set.seed(99)
    post=nft(x.train, times, delta, K=0)
    pred=predict(post, x.train, XPtr=TRUE, seed=21)
    print(Cindex(pred$logt.test.mean, times, delta))

## End(Not run)