On nonparametric likelihood ratio estimation of survival probabilities for censored data

doi:10.1016/0167-7152(94)00210-Y

Statistics & Probability Letters

Volume 25, Issue 2, 1 November 1995, Pages 95-104

https://doi.org/10.1016/0167-7152(94)00210-Y Get rights and content

Abstract

Thomas and Grunkemeier (1975) proposed a nonparametric likelihood ratio method for interval estimation of survival probabilities for randomly censored data. Their method always produces confidence intervals inside [0,1] and has a better performance than the normal approximation method based on the Kaplan-Meier estimate and Greenwood's formula. In this note we show that the likelihood ratio used by Thomas and Grunkemeier (1978) is a “genuine” nonparametric likelihood ratio. That is, it can be derived by considering the parameter space of all survival functions. This property is not shared by many existing empirical likelihood methods. We also note that this result is not a direct consequence of the fact proved by Kaplan and Meier (1958) that the maximum of the likelihood over the space of all survival functions is achieved in the subspace of all discrete survival functions supported on the observed uncensored lifetimes. Another objective of this note is to point out that the likelihood ratio approach can also be used to draw joint inferences on any finite number of probabilities, and test goodness of fit of a given survival function for censored data. A rigorous derivation for the limiting distribution of the likelihood ratio is also provided.

References (17)

N.E. Breslow et al.
A large sample study of the life table and product limit estimates under random censorship
Ann. Statist.
(1974)
D.R. Cox et al.
Analysis of Survival Data
(1984)
F. Critchley et al.
Interval estimation based on the profile likelihood: Strong Lagrangian theory with applications to discrimination
Biometrika
(1988)
T.J. DiCiccio et al.
Empirical likelihood is Bartlett-correctable
Ann. Statist.
(1991)
T.R. Fleming et al.
Counting Processes and Survival Analysis
(1991)
R.D. Gill
Censoring and Stochastic Integrals
P. Hall
Pseudo-likelihood theory for empirical likelihood
Ann. Statist.
(1990)
E. Kaplan et al.
Nonparametric estimation from incomplete observations
J. Amer. Statist. Assoc.
(1958)

There are more references available in the full text version of this article.

Cited by (63)

Simultaneous confidence bands for survival functions from twice censorship
2022, Statistics and Probability Letters
Citation Excerpt :
Owen (1988, 1990) gave the first theoretical treatment of the nonparametric LR method. For RC data related investigations, see Li (1995), Li et al. (1996), Einmahl and McKeague (1999), Subramanian (2016), among others. TC related LR investigations, however, are completely absent.
When a failure time observation is right censored, it is not observed beyond a random right threshold. When a possibly right censored observation is further susceptible to a random left threshold, a twice censored observation results. Patilea and Rolin have proposed product limit estimators for survival functions from twice censored data. Simultaneous confidence bands (SCBs) for survival functions from twice censored data are constructed in a way that mimics the approach that Hollander, McKeague and Yang pursued for random right censoring. Their nonparametric likelihood ratio function is adjusted, providing the basis for constructing the SCBs. The critical value needed for the SCBs is obtained using the bootstrap, for which asymptotic justification is provided. The SCBs align nicely as “neighborhoods” of the Patilea–Rolin nonparametric survival function estimator, in much the same way the likelihood ratio SCBs under random censoring are the “neighborhoods” of the Kaplan–Meier estimator. A simulation study supports the effectiveness of the proposed method. An illustration is given using synthetic data.
Bootstrap likelihood ratio confidence bands for survival functions under random censorship and its semiparametric extension
2016, Journal of Multivariate Analysis
Simultaneous confidence bands for survival functions, from randomly right censored data, can be computed by inverting likelihood ratio functions based on appropriate thresholds. Sometimes, however, the requisite asymptotic distributions are intractable, or thresholds based on Brownian bridge approximations are not easy to obtain when simultaneous confidence bands over only sub-regions are possible or desired. We obtain the thresholds by bootstrapping (i) a nonparametric likelihood ratio function via censored data bootstrap and (ii) a semiparametric adjusted likelihood ratio function via a two-stage bootstrap that utilizes a model for the second stage. These two scenarios are grounded respectively in standard random censorship and its semiparametric extension introduced by Dikta. The two bootstraps, which are different in the way resampling is done, are shown to have asymptotic validity. The respective confidence bands are neighborhoods of the well-known Kaplan–Meier estimator and the more recently developed Dikta’s semiparametric counterpart. As evidenced by a validation study, both types of confidence bands provide approximately correct coverage. The model-based confidence bands, however, are tighter than the nonparametric ones. Two sensitivity studies reveal that the model-based method performs well when standard binary regression models are fitted, indicating its robustness to misspecification as well as its practical applicability. An illustration is given using real data.
Empirical likelihood confidence regions for one- or two- samples with doubly censored data
2016, Computational Statistics and Data Analysis
Citation Excerpt :
Another approach to handling incomplete sample is to construct likelihood ratio directly from incomplete data. This method can be traced back to Thomas and Grunkemeier (1975) who used the nonparametric likelihood ratio method to construct confidence intervals for survival probabilities under random censorship (see also Li (1995)). For right-censored data, Li and Van Keilegom (2002) obtained confidence bands for conditional survival and quantile functions using the nonparametric likelihood ratio approach.
The purpose is to propose a new EM algorithm for doubly censored data subject to nonparametric moment constraints. Empirical likelihood confidence regions are constructed for one- or two- samples of doubly censored data. It is shown that the corresponding empirical likelihood ratio converges to a standard chi-square random variable. Simulations are carried out to assess the finite-sample performance of the proposed method. For illustration purpose, the proposed method is applied to a real data set with two samples.
Empirical likelihood inference for the odds ratio of two survival functions under right censoring
2015, Statistics and Probability Letters
Smoothed empirical likelihood for ROC curves with censored data
2012, Journal of Multivariate Analysis
The receiver operating characteristic (ROC) curve is an attractive basis for the comparison of distribution functions between two populations. In this paper, we apply the censored empirical likelihood method with kernel smoothing to investigate the ROC curve. It is shown that the smoothed empirical likelihood ratio converges to a chi-square distribution, which is the well-known Wilks theorem. We also propose a bootstrap procedure for obtaining the censored empirical likelihood confidence band for the ROC curve. The performance of the proposed empirical likelihood method is illustrated through extensive simulation studies in terms of coverage probability and average length of confidence intervals.
Model-based likelihood ratio confidence intervals for survival functions
2012, Statistics and Probability Letters
We introduce an adjusted likelihood ratio procedure for computing pointwise confidence intervals for survival functions from censored data. The test statistic, scaled by a ratio of two variance quantities, is shown to converge to a chi-squared distribution with one degree of freedom. The confidence intervals are seen to be a neighborhood of a semiparametric survival function estimator and are shown to have correct empirical coverage. Numerical studies also indicate that the proposed intervals have smaller estimated mean lengths in comparison to the ones that are produced as a neighborhood of the Kaplan–Meier estimator. We illustrate our method using a lung cancer data set.

View all citing articles on Scopus

^☆: This work was supported in part by funds provided by The University of North Carolina at Charlotte.

View full text

On nonparametric likelihood ratio estimation of survival probabilities for censored data☆

Abstract

A large sample study of the life table and product limit estimates under random censorship

Ann. Statist.

Analysis of Survival Data

Interval estimation based on the profile likelihood: Strong Lagrangian theory with applications to discrimination

Biometrika

Empirical likelihood is Bartlett-correctable

Ann. Statist.

Counting Processes and Survival Analysis

Censoring and Stochastic Integrals

Pseudo-likelihood theory for empirical likelihood

Ann. Statist.

Nonparametric estimation from incomplete observations

J. Amer. Statist. Assoc.