Efficient nonparametric confidence bands for receiver operating-characteristic curves

Martínez–Camblor, Pablo; Pérez-Fernández, Sonia; Corral, Norberto

doi:10.1177/0962280216672490

Cited by 8 publications

(12 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this package three different techniques dealing with the ROC curve itself have been computed. Namely, one parametric assuming the binormal model (Demidenko (2012)) and two non-parametric have been included (Jensen et al (2000) and Martínez-Camblor et al (2018)):…”

Section: Pairpointscoordinatesmentioning

confidence: 99%

“…The bootstrap method has been applied and the confidence bands are truncated making the lower-band being inside the (0, 0.95) interval and the upper-band within (0.05, 1). • Martínez-Camblor et al (2018) method approximates the distribution of the following pivotal function by a smoothed bootstrap method:…”

Section: Pairpointscoordinatesmentioning

confidence: 99%

“…method Method used. One of "PSN" (Martínez-Camblor et al, 2018), "JMS" (Jensen et al, 2000) or "DEK" (Demidenko, 2012…”

Section: Input Parameters Grocmentioning

confidence: 99%

“…The code computed, using the metaROC function, is listed below: • confidence bands construction by three different methods: two of them are non-parametric (Jensen et al (2000) and Martínez-Camblor et al (2018)) and the other one is based on the binormal model (Demidenko (2012) In spite of the popularity of R packages about ROC curves dealing with some of the most important analyses related to this tool, nsROC includes some algorithms which had not been computed to date in order to address some of those standard analyses (such as time-dependent ROC curve estimation and comparison between curves) and others totally new such as the generalized ROC curve estimation and non-parametric procedure for meta-analysis. Any of these particular techniques had been addressed earlier, excluding the usual estimation of the curve, the weighted Kaplan-Meier method to deal with the presence of censored data in time-dependent ROC curves estimation, and the Venkatraman and DeLong approaches to compare diagnostic accuracies of two tests.…”

Section: Interleukin 6 (Il6) Datamentioning

confidence: 99%

“…The interested reader can consult the monographs of Zhou et al (2002) and Pepe (2003) for an extensive review of the topic. There are also a number of papers dealing with some problems related to ROC curve such as the usual ROC curve point estimation (see Gonçalvez et al (2014) for a recent overview) from both parametric and non-parametric approaches, even considering Bayesian methods as an alternative to the maximum likelihood principle; or the curve interval estimation (confidence bands construction) also using both parametric (Demidenko, 2012) and non-parametric techniques (Jensen et al (2000), Horváth et al (2008) and Martínez-Camblor et al (2018)). Furthermore, the ROC curve procedure has been extended to other situations where the outcome is not binary.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

nsROC: An R package for Non-Standard ROC Curve Analysis

Pérez-Fernández¹,

Martínez–Camblor²,

Filzmoser³

et al. 2019

The R Journal

Self Cite

View full text Add to dashboard Cite

The receiver operating characteristic (ROC) curve is a graphical method which has become standard in the analysis of diagnostic markers, that is, in the study of the classification ability of a numerical variable. Most of the commercial statistical software provide routines for the standard ROC curve analysis. Of course, there are also many R packages dealing with the ROC estimation as well as other related problems. In this work we introduce the nsROC package which incorporates some new ROC curve procedures. Particularly: ROC curve comparison based on general distances among functions for both paired and unpaired designs; efficient confidence bands construction; a generalization of the curve considering different classification subsets than the one involved in the classical definition of the ROC curve; a procedure to deal with censored data in cumulative-dynamic ROC curve estimation for time-to-event outcomes; and a non-parametric ROC curve method for meta-analysis. This is the only R package which implements these particular procedures. Estimation Non-standard ROC curve estimationAs mentioned previously, an ROC curve is a graphical method which displays the sensitivity (Se) versus the complementary of the specificity (1-Sp) for all possible thresholds of the considered marker.Although different parametric and semi-parametric estimators for the ROC curve have been studied, in our package the empirical estimator, based on replacing the involved unknown distribution functions with their respective empirical cumulative distribution functions,F, has been considered. Hence, the implemented ROC curve estimator isAbstract The increase in life expectancy followed by the burden of chronic diseases contributes to disability at older ages. The estimation of how much chronic conditions contribute to disability can be useful to develop public health strategies to reduce the burden. This paper introduces the R package addhaz, which is based on the attribution method (Nusselder and Looman, 2004) to partition disability into the additive contributions of diseases using cross-sectional data. The R package includes tools to fit the additive hazard model, the core of the attribution method, to binary and multinomial outcomes. The models are fitted by maximizing the binomial and multinomial log-likelihood functions using constrained optimization. Wald and bootstrap confidence intervals can be obtained for the parameter estimates. Also, the contribution of diseases to the disability prevalence and their bootstrap confidence intervals can be estimated. An additional feature is the possibility to use parallel computing to obtain the bootstrap confidence intervals. In this manuscript, we illustrate the use of addhaz with several examples for the binomial and multinomial models, using the data from the Brazilian National Health Survey, 2013.

show abstract

Section: Pairpointscoordinatesmentioning

confidence: 99%

Section: Pairpointscoordinatesmentioning

confidence: 99%

“…method Method used. One of "PSN" (Martínez-Camblor et al, 2018), "JMS" (Jensen et al, 2000) or "DEK" (Demidenko, 2012…”

Section: Input Parameters Grocmentioning

confidence: 99%

Section: Interleukin 6 (Il6) Datamentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

nsROC: An R package for Non-Standard ROC Curve Analysis

Pérez-Fernández¹,

Martínez–Camblor²,

Filzmoser³

et al. 2019

The R Journal

Self Cite

View full text Add to dashboard Cite

show abstract

ROC curves for clinical prediction models part 1. ROC plots showed no added value above the AUC when evaluating the performance of clinical prediction models

Verbakel

Steyerberg

Uno

et al. 2020

Journal of Clinical Epidemiology

View full text Add to dashboard Cite

Objectives: Receiver operating characteristic (ROC) curves show how well a risk prediction model discriminates between patients with and without a condition. We aim to investigate how ROC curves are presented in the literature and discuss and illustrate their potential limitations.Study Design and Setting: We conducted a pragmatic literature review of contemporary publications that externally validated clinical prediction models. We illustrated limitations of ROC curves using a testicular cancer case study and simulated data.Results: Of 86 identified prediction modeling studies, 52 (60%) presented ROC curves without thresholds and one (1%) presented an ROC curve with only a few thresholds. We illustrate that ROC curves in their standard form withhold threshold information have an unstable shape even for the same area under the curve (AUC) and are problematic for comparing model performance conditional on threshold. We compare ROC curves with classification plots, which show sensitivity and specificity conditional on risk thresholds.Conclusion: ROC curves do not offer more information than the AUC to indicate discriminative ability. To assess the model's performance for decision-making, results should be provided conditional on risk thresholds. Therefore, if discriminatory ability must be visualized, classification plots are attractive.

show abstract

Analysis of endoplasmic reticulum stress-related gene signature for the prognosis and pattern in diffuse large B cell lymphoma

Zhang,

Lin,

et al. 2023

Sci Rep

View full text Add to dashboard Cite

Diffuse large B-cell lymphoma (DLBCL) is the most common lymphoma in adults. This study aimed to determine the prognostic significance of endoplasmic reticulum (ER) stress-related genes in DLBCL. ER stress-related genes were obtained from the molecular signatures database. Gene expression data and clinical outcomes from the gene expression omnibus and TCGA datasets were collected, and differentially expressed genes (DEGs) were screened out. Gene ontology enrichment analysis, the kyoto encyclopaedia of genes and genomes pathway analysis, and geneset enrichment analysis were used to analyse the possible biological function of ER stress-related DEGs in DLBCL. Protein–protein interaction network construction using the STRING online and hub genes were identified by cytoHubba on Cytoscape software. The significant prognosis-related genes were screened, and the differential expression was validated. The immune microenvironment assessment of significant genes were evaluated. Next, the nomogram was built using univariate and multivariate Cox regression analysis. 26 ER stress-related DEGs were screened. Functional enrichment analysis showed them to be involved in the regulation of the endoplasmic reticulum mainly. NUPR1 and TRIB3 were identified as the most significant prognostic-related genes by comparison with the GSE10846, GSE11318, and TCGA datasets. NUPR1 was correlated with a good prognosis and immune infiltration in DLBCL; on the other hand, high expression of TRIB3 significantly correlated with a poor prognosis, which was an independent prognostic factor for DLBCL. In summary, we identified NUPR1 and TRIB3 as critical ER stress-related genes in DLBCL. NUPR1 might be involved in immune infiltration in DLBCL, and TRIB3 might serve as a potential therapeutic target and prognostic factor in DLBCL.

show abstract

Efficient nonparametric confidence bands for receiver operating-characteristic curves

Cited by 8 publications

References 31 publications

nsROC: An R package for Non-Standard ROC Curve Analysis

nsROC: An R package for Non-Standard ROC Curve Analysis

ROC curves for clinical prediction models part 1. ROC plots showed no added value above the AUC when evaluating the performance of clinical prediction models

Analysis of endoplasmic reticulum stress-related gene signature for the prognosis and pattern in diffuse large B cell lymphoma

Contact Info

Product

Resources

About