Linear Combinations of Multiple Diagnostic Markers

Su, John Q.; Liu, Jun S.

doi:10.1080/01621459.1993.10476417

Cited by 211 publications

(174 citation statements)

References 13 publications

Supporting

Mentioning

171

Contrasting

Order By: Relevance

“…First, combining multiple variables and conditions (e.g., AREA/FC, AREA/ FO and FD/RO) using standard statistical tools, such as Linear Discriminant Analysis (Su and Liu, 1993) or multi-dimensional classification algorithms, should improve the discrimination quality. Second, assuming a higher discrimination quality, it should be possible to use a dichotomy search to determine TA with higher precision.…”

Section: Discussionmentioning

confidence: 99%

“…To assess the robustness of the protocol, we plotted the TPR versus the FPR (yielding the Receiver Operating Characteristic or ROC curve). The robustness of the protocol could then be quantified by the Area Under the ROC Curve (AUC), with AUC = 1 for perfect discrimination and AUC = 0.5 for indifferent discrimination (Su and Liu, 1993).…”

Section: Assessment Of the Robustness Of The Protocolmentioning

confidence: 99%

See 1 more Smart Citation

Robust evaluation of time since awakening using force platform posturography

Pham

Mello

Narciso

et al. 2014

Rev. Bras. Eng. Bioméd.

View full text Add to dashboard Cite

Introduction: Sleepiness is responsible for a considerable proportion of traffi c accidents. It is thus an important traffi c safety issue to fi nd a robust, objective and practical way to estimate the amount of time a person has been awake. To attempt to meet this goal, we investigated the relationship between sleepiness and posture control. Methods: Subjects were kept awake for 36 hours and posturographic data during quiet standing were collected every two hours by means of a force platform. The standing surface (rigid surface or foam surface) and visual (eyes open or eyes closed) conditions were manipulated. Results: In the more challenging conditions (with foam surface and/or eyes closed), the body sway variables derived from the center of the pressure measurement increased signifi cantly when time since awakening became greater than 21 h in almost all subjects. Conclusion: Based on this result, we propose a practical protocol that could robustly assess whether time since awakening was greater than 21 h.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Assessment Of the Robustness Of The Protocolmentioning

confidence: 99%

Robust evaluation of time since awakening using force platform posturography

Pham

Mello

Narciso

et al. 2014

Rev. Bras. Eng. Bioméd.

View full text Add to dashboard Cite

show abstract

“…Then, effect sizes were introduced in a way that each set of covariates achieved an (added) performance approximately in the magnitude of a pre-specified area under the receiver-operating characteristic (ROC) curve (AUC) value. As a reference we used the theoretical relationship [15]: …”

Section: Methodsmentioning

confidence: 99%

Assessment of predictive performance in incomplete data by combining internal validation and multiple imputation

Wahl

Boulesteix

Zierer

et al. 2016

BMC Med Res Methodol

View full text Add to dashboard Cite

BackgroundMissing values are a frequent issue in human studies. In many situations, multiple imputation (MI) is an appropriate missing data handling strategy, whereby missing values are imputed multiple times, the analysis is performed in every imputed data set, and the obtained estimates are pooled. If the aim is to estimate (added) predictive performance measures, such as (change in) the area under the receiver-operating characteristic curve (AUC), internal validation strategies become desirable in order to correct for optimism. It is not fully understood how internal validation should be combined with multiple imputation.MethodsIn a comprehensive simulation study and in a real data set based on blood markers as predictors for mortality, we compare three combination strategies: Val-MI, internal validation followed by MI on the training and test parts separately, MI-Val, MI on the full data set followed by internal validation, and MI(-y)-Val, MI on the full data set omitting the outcome followed by internal validation. Different validation strategies, including bootstrap und cross-validation, different (added) performance measures, and various data characteristics are considered, and the strategies are evaluated with regard to bias and mean squared error of the obtained performance estimates. In addition, we elaborate on the number of resamples and imputations to be used, and adopt a strategy for confidence interval construction to incomplete data.ResultsInternal validation is essential in order to avoid optimism, with the bootstrap 0.632+ estimate representing a reliable method to correct for optimism. While estimates obtained by MI-Val are optimistically biased, those obtained by MI(-y)-Val tend to be pessimistic in the presence of a true underlying effect. Val-MI provides largely unbiased estimates, with a slight pessimistic bias with increasing true effect size, number of covariates and decreasing sample size. In Val-MI, accuracy of the estimate is more strongly improved by increasing the number of bootstrap draws rather than the number of imputations. With a simple integrated approach, valid confidence intervals for performance estimates can be obtained.ConclusionsWhen prognostic models are developed on incomplete data, Val-MI represents a valid strategy to obtain estimates of predictive performance measures.Electronic supplementary materialThe online version of this article (doi:10.1186/s12874-016-0239-7) contains supplementary material, which is available to authorized users.

show abstract

“…Usually this problem is solved adding supplementary conditions or hypoteses to variables X 1 and X 2 [2-8]. In [9,10] there are two comprehensive surveys.…”

Section: Background and Previous Resultsmentioning

confidence: 99%

Tools to identify linear combination of prognostic factors which maximizes area under receiver operator curve

Todor

Săplăcan³

2014

J Clin Bioinformatics

View full text Add to dashboard Cite

BackgroundThe linear combination of variables is an attractive method in many medical analyses targeting a score to classify patients. In the case of ROC curves the most popular problem is to identify the linear combination which maximizes area under curve (AUC). This problem is complete closed when normality assumptions are met. With no assumption of normality search algorithm are avoided because it is accepted that we have to evaluate AUC nd times where n is the number of distinct observation and d is the number of variables.MethodsFor d = 2, using particularities of AUC formula, we described an algorithm which lowered the number of evaluations of AUC from n2 to n(n-1) + 1. For d > 2 our proposed solution is an approximate method by considering equidistant points on the unit sphere in Rd where we evaluate AUC.ResultsThe algorithms were applied to data from our lab to predict response of treatment by a set of molecular markers in cervical cancers patients. In order to evaluate the strength of our algorithms a simulation was added.ConclusionsIn the case of no normality presented algorithms are feasible. For many variables computation time could be increased but acceptable.

show abstract

Linear Combinations of Multiple Diagnostic Markers

Cited by 211 publications

References 13 publications

Robust evaluation of time since awakening using force platform posturography

Robust evaluation of time since awakening using force platform posturography

Assessment of predictive performance in incomplete data by combining internal validation and multiple imputation

Tools to identify linear combination of prognostic factors which maximizes area under receiver operator curve

Contact Info

Product

Resources

About