Point Estimates of Test Sensitivity and Specificity from Sample Means and Variances

In a wide variety of biomedical and clinical research studies, sample statistics from diagnostic marker measurements are presented as a means of distinguishing between two populations, such as with and without disease. Intuitively, a larger difference between the mean values of a marker for the two populations, and a smaller spread of values within each population, should lead to more reliable classification rules based on this marker. We formalize this intuitive notion by deriving practical, new, closed-form expressions for the sensitivity and specificity of three different discriminant tests defined in terms of the sample means and standard deviations of diagnostic marker measurements. The three discriminant tests evaluated are based, respectively, on the Euclidean distance and the Mahalanobis distance between means, and a likelihood ratio analysis. Expressions for the effects of measurement error are also presented. Our final expressions assume that the diagnostic markers follow independent normal distributions for the two populations, although it will be clear that other known distributions may be similarly analyzed. We then discuss applications drawn from the medical literature, although the formalism is clearly not restricted to that application.     

Model Checks in Inverse Regression Models with Convolution‐Type Operators

Abstract. We consider the problem of testing parametric assumptions in an inverse regression model with a convolution‐type operator. An L ₂ ‐type goodness‐of‐fit test is proposed which compares the distance between a parametric and a non‐parametric estimate of the regression function. Asymptotic normality of the corresponding test statistic is shown under the null hypothesis and under a general non‐parametric alternative with different rates of convergence in both cases. The feasibility of the proposed test is demonstrated by means of a small simulation study. In particular, the power of the test against certain types of alternative is investigated. Finally, an empirical example is provided, in which the proposed methods are applied to the determination of the shape of the luminosity profile of the elliptical galaxy NGC 5017.     

Comparing diagnostic tests: test of hypothesis for likelihood ratios

Likelihood ratios (LRs) are used to characterize the efficiency of diagnostic tests. In this paper, we use the classical weighted least squares (CWLS) test procedure, which was originally used for testing the homogeneity of relative risks, for comparing the LRs of two or more binary diagnostic tests. We compare the performance of this method with the relative diagnostic likelihood ratio (rDLR) method and the diagnostic likelihood ratio regression (DLRReg) approach in terms of size and power, and we observe that the performances of CWLS and rDLR are the same when used to compare two diagnostic tests, while DLRReg method has higher type I error rates and powers. We also examine the performances of the CWLS and DLRReg methods for comparing three diagnostic tests in various sample size and prevalence combinations. On the basis of Monte Carlo simulations, we conclude that all of the tests are generally conservative and have low power, especially in settings of small sample size and low prevalence.     

Comparison of procedures for fitting the autoregressive order of a vector error correction model

This paper investigates the lag length selection problem of a vector error correction model by using a convergent information criterion and tools based on the Box–Pierce methodology recently proposed in the literature. The performances of these approaches for selecting the optimal lag length are compared via Monte Carlo experiments. The effects of misspecified deterministic trend or cointegrating rank on the lag length selection are studied. Noting that processes often exhibit nonlinearities, the cases of iid and conditionally heteroscedastic errors will be considered. Strategies that can avoid misleading situations are proposed.     

Monte Carlo comparison of five exponentiality tests using different entropy estimates

The paper studies five entropy tests of exponentiality using five statistics based on different entropy estimates. Critical values for various sample sizes determined by means of Monte Carlo simulations are presented for each of the test statistics. By simulation, we compare the power of these five tests for various alternatives and sample sizes.     

An adaptive two-sample location-scale test of lepage type for symmetric distributions

For the two-sample location and scale problem we propose an adaptive test which is based on so called Lepage type tests. The well known test of Lepage (1971) is a combination of the Wilcoxon test for location alternatives and the Ansari-Bradley test for scale alternatives and it behaves well for symmetric and medium-tailed distributions. For the cae of short-, medium- and long-tailed distributions we replace the Wilcoxon test and the .Ansari-Bradley test by suitable other two-sample tests for location and scale, respectively, in oder to get higher power than the classical Lepage test for such distribotions. These tests here are called Lepage type tests. in practice, however, we generally have no clear idea about the distribution having generated our data. Thus, an adaptive test should be applied which takes the the given data set inio consideration. The proposed adaptive test is based on the concept of Hogg (1974), i.e., first, to classify the unknown symmetric distribution function with respect to a measure for tailweight and second, to apply an appropriate Lepage type test for this classified type of distribution. We compare the adaptive test with the three Lepage type tests in the adaptive scheme and with the classical Lepage test as well as with other parametric and nonparametric tests. The power comparison is carried out via Monte Carlo simulation. It is shown that the adaptive test is the best one for the broad class of distributions considered.     

Bayesian Network Structure Learning with Permutation Tests

In literature there are several studies on the performance of Bayesian network structure learning algorithms. The focus of these studies is almost always the heuristics the learning algorithms are based on, i.e., the maximization algorithms (in score-based algorithms) or the techniques for learning the dependencies of each variable (in constraint-based algorithms). In this article, we investigate how the use of permutation tests instead of parametric ones affects the performance of Bayesian network structure learning from discrete data. Shrinkage tests are also covered to provide a broad overview of the techniques developed in current literature.     

Dependency and truncated forms of combinations in multivariate combination-based permutation tests and ordered categorical variables

ABSTRACT

Quite an important problem usually occurs in several multi-dimensional hypotheses testing problems when variables are correlated. In this framework the non-parametric combination (NPC) of a finite number of dependent permutation tests is suitable to cover almost all real situations of practical interest since the dependence relations among partial tests are implicitly captured by the combining procedure itself without the need to specify them [Pesarin F, Salmaso L. Permutation tests for complex data: theory, applications and software. Chichester: Wiley; 2010a]. An open problem related to NPC-based tests is the impact of the dependency structure on combined tests, especially in the presence of categorical variables. This paper’s goal is firstly to investigate the impact of the dependency structure on the possible significance of combined tests in cases of ordered categorical responses using Monte Carlo simulations, then to propose some specific procedures aimed at improving the power of multivariate combination-based permutation tests. The results show that an increasing level of correlation/association among responses negatively affects the power of combination-based multivariate permutation tests. The application of special forms of combination functions based on the truncated product method [Zaykin DV, Zhivotovsky LA, Westfall PH, Weir BS. Truncated product method for combining p-values. Genet Epidemiol. 2002;22:170–185; Dudbridge F, Koeleman BPC. Rank truncated product of p-values, with application to genomewide association scans. Genet Epidemiol. 2003;25:360–366] or on Liptak combination allowed us, using Monte Carlo simulations, to demonstrate the possibility of mitigating the negative effect on power of combination-based multivariate permutation tests produced by an increasing level of correlation/association among responses.     

Effect of tuberculosis on immune restoration among HIV-infected patients receiving antiretroviral therapy

In this article, time to immune recovery during antiretroviral therapy was estimated and compared between HIV-infected children with and without tuberculosis (TB). CD4?T-cell restoration was used as a criterion for determining immune recovery. The median residual lifetime function, which is more intuitive and robust compared to the frequently used measures of lifetime data, was used to estimate time to CD4?T-cell restoration. The median residual lifetime is not influenced by extreme observations and heavy-tailed distributions which are commonly encountered in clinical studies. Permutation-based methods were used to compare the CD4?T-cell restoration times between the two groups of patients. Our results indicate that children with TB had uniformly higher median residual lifetimes to immune recovery compared to those without TB. Although TB was associated with slower CD4?T-cell restoration, the differences between the restoration times of the two groups were not statistically significant.