Non-strange weird resampling for complex survival data

Dobler, Dennis; Beyersmann, Jan; Pauly, Markus

doi:10.1093/biomet/asx026

Cited by 16 publications

(17 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…But in cases, where this is so, the deviance is only very small. The phenomenon, that standard normal multipliers yield a worse performance than those with skewness equal to one, is in line with the findings in a revised version of Dobler et al (2015) where also heuristic theoretic arguments for a second-order correctness of both superior resampling procedures are provided. As there is, all in all, not much of a difference between the simulated coverage probabilities of the centered unit Poisson wild bootstrap and the weird bootstrap, we only focus on the results of the Poisson choice.…”

Section: Small Sample Behavioursupporting

confidence: 78%

“…This resampling scheme corresponds to choosing ξ j i`1 " BinpY pX i q, maxp1, Y pX i qq´1q, where X i is the censoring or event time of individual i, whichever comes first. This is a particular choice of the data-dependent multiplier bootstrap of Dobler et al (2015). In their article, heuristic arguments for the second order correctness under absolute continuity of the data have shown that centered unit Poisson variates and weird bootstrap multipliers perform favorably in comparison to standard normal wild bootstrap weights.…”

Section: Now An Appropriate Wild Bootstrap Version Ofmentioning

confidence: 99%

See 1 more Smart Citation

A discontinuity adjustment for subdistribution function confidence bands applied to right-censored competing risks data

Dobler¹

2017

Electron. J. Statist.

View full text Add to dashboard Cite

The wild bootstrap is the resampling method of choice in survival analytic applications. Theoretic justifications rely on the assumption of existing intensity functions which is equivalent to an exclusion of ties among the event times. However, such ties are omnipresent in practical studies. It turns out that the wild bootstrap should only be applied in a modified manner that corrects for altered limit variances and emerging dependencies. This again ensures the asymptotic exactness of inferential procedures. An analogous necessity is the use of the Greenwood-type variance estimator for Nelson-Aalen estimators which is particularly preferred in tied data regimes. All theoretic arguments are transferred to bootstrapping Aalen-Johansen estimators for cumulative incidence functions in competing risks. An extensive simulation study as well as an application to real competing risks data of male intensive care unit patients suffering from pneumonia illustrate the practicability of the proposed technique.where the last equality follows from the orthogonality of all involved single martingales. Inserting the above representation for the optional variation process yieldsIn a last step, we replace the unknown increments of the cumulative hazard function ∆A 1 puq with Nelson-Aalen increments ∆ p A 1 puq " Y puq´1∆N 1 puq and obtain the Greenwood-type variance estimator

show abstract

Section: Small Sample Behavioursupporting

confidence: 78%

Section: Now An Appropriate Wild Bootstrap Version Ofmentioning

confidence: 99%

A discontinuity adjustment for subdistribution function confidence bands applied to right-censored competing risks data

Dobler¹

2017

Electron. J. Statist.

View full text Add to dashboard Cite

show abstract

“…Since the minimally selected test statistics in our paper do not have a nice martingale representation form and are not normally distributed, the wild bootstrap method of Lin (1997) is not applicable to our setting. The weird bootstrap and its extensions (Andersen et al, 2012; Dobler et al, 2017) consider the Nelson–Aalen estimator, and replace the counting process increments at time t with binomial random variables with size Y ( t ) and success probability d Â ( t ), where Y ( t ) is the size of risk set at t and d Â ( t ) is the empirical hazard function at t . It is possible that we could generate a weird bootstrap replication through this process – under the null of quasi-independence – and use it to form a reference distribution for our test statistics.…”

Section: Discussionmentioning

confidence: 99%

Permutation tests for general dependent truncation

Chiou

Qian

Mormino

et al. 2018

Computational Statistics & Data Analysis

View full text Add to dashboard Cite

Truncated survival data arise when the event time is observed only if it falls within a subject-specific region, known as the truncation set. Left-truncated data arise when there is delayed entry into a study, such that subjects are included only if their event time exceeds some other time. Quasi-independence of truncation and failure refers to factorization of their joint density in the observable region. Under quasi-independence, standard methods for survival data such as the Kaplan–Meier estimator and Cox regression can be applied after simple adjustments to the risk sets. Unlike the requisite assumption of independent censoring, quasi-independence can be tested, e.g., using a conditional Kendall’s tau test. Current methods for testing for quasi-independence are powerful for monotone alternatives. Nonetheless, it is essential to detect any kind of deviation from quasi-independence so as not to report a biased Kaplan–Meier estimator or regression effect, which would arise from applying the simple risk set adjustment when dependence holds. Nonparametric, minimum p-value tests that are powerful against non-monotone alternatives are developed to offer protection against erroneous assumptions of quasi-independence. The use of conditional and unconditional methods of permutation for evaluation of the proposed tests are investigated in simulation studies. The proposed tests are applied to a study on the cognitive and functional decline in aging.

show abstract

“…() as well as Dobler et al. () recently gave a thorough mathematical treatment of this approach also allowing for independent left‐truncation and general multipliers. Extensions to more complex multistate models are, however, rare in the current literature.…”

Section: Introductionmentioning

confidence: 99%

A Wild Bootstrap Approach for the Aalen–Johansen Estimator

Bluhmki¹,

Schmoor

Dobler³

et al. 2018

Biometrics

View full text Add to dashboard Cite

We suggest a wild bootstrap resampling technique for nonparametric inference on transition probabilities in a general time-inhomogeneous Markov multistate model. We first approximate the limiting distribution of the Nelson-Aalen estimator by repeatedly generating standard normal wild bootstrap variates, while the data is kept fixed. Next, a transformation using a functional delta method argument is applied. The approach is conceptually easier than direct resampling for the transition probabilities. It is used to investigate a non-standard time-to-event outcome, currently being alive without immunosuppressive treatment, with data from a recent study of prophylactic treatment in allogeneic transplanted leukemia patients. Due to non-monotonic outcome probabilities in time, neither standard survival nor competing risks techniques apply, which highlights the need for the present methodology. Finite sample performance of time-simultaneous confidence bands for the outcome probabilities is assessed in an extensive simulation study motivated by the clinical trial data. Example code is provided in the web-based Supplementary Materials.

show abstract

Non-strange weird resampling for complex survival data

Cited by 16 publications

References 30 publications

A discontinuity adjustment for subdistribution function confidence bands applied to right-censored competing risks data

A discontinuity adjustment for subdistribution function confidence bands applied to right-censored competing risks data

Permutation tests for general dependent truncation

A Wild Bootstrap Approach for the Aalen–Johansen Estimator

Contact Info

Product

Resources

About