摘要:

Pharmacokinetic (PK) models describe how the concentrations of a drug and its metabolite vary with time. Population PK models identify and quantify sources of between-individual variability in observed concentrations. Crucial to this aim is the identification of those covariates (i.e., individual-specific characteristics) responsible for explaining the variability. In this article we discuss how covariate modeling can be carried out for population PK models. We argue that the importance of a particular covariate can be discussed only with reference to the specific use for which the model is intended. Covariate modeling is important in population PK studies as it aids in determining dosage recommendations for specific covariate-defined populations. We describe a Bayesian predictive procedure that places covariate modeling in the context of dosage determination. In problems such as these it is crucial to incorporate relevant prior information. For covariateselectionwe extend the approach of George and McCulloch. The approaches utilize Markov chain Monte Carlo techniques. The methods are illustrated using population PK data from a study of the antibiotic vancomycin in babies. These data are sparse, with just 180 concentrations from 37 babies. Eight covariates are available, from which we construct a covariate model.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476961

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

We propose a likelihood-based method to analyze the causal effect of partial compliance (i.e., unplanned partial exposure to treatment or placebo) in the LRC-CPPT data, a prevention trial with long term follow-up previously analyzed by Efron and Feldman. Initially, we construct ordered compliance categories and dichotomize response. Assuming increased exposure to cholestyramine does not increase cholesterol, we estimate exposure—response curves in different compliance subsets. Subjects in different arms with similar levels of compliance to the assignment may have a different placebo prognosis (i.e., success probability under a possible zero exposure level). The sole assumption that the placebo group reflects response to zero exposure for the treatment group as a whole allows estimation of a causal parameter in a special case only. When a single parameter represents the association between responses to possible treatment exposures and treatment compliance, simple estimates are derived for a set of causal parameters. The example is analyzed in detail, and more general applicability and extensions of the method are discussed.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476962

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

This article considers the estimation of the rate of vertical human immunodeficiency virus (HIV) transmission and the pediatric acquired immune deficiency syndrome (AIDS) incubation period using data from birth cohort studies. Standard methods of analysis, which ignore children whose infection status is not established, may lead to biased estimates. Methods based on modeling the disappearance of HIV antibody or appearance of virus are inefficient, as they essentially rely on a single variable. We describe an alternative model that takes into account clinical, immunological, and virological data. Maximum likelihood estimates of rate of vertical transmission, sensitivity of virus tests, pediatric AIDS incubation period, and age distribution at antibody loss are readily obtained by an EM algorithm. The method was applied to data from the European Collaborative Study and revealed evidence of temporal changes in the transmission rate and AIDS incubation period. This new approach, with possible modification, should allow efficient analysis of data from randomized controlled trials of interventions to reduce vertical transmission. It may also be applicable to other vertically transmissible infections.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476963

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

This article considers the problem of identifying the irregular boundary of a magnetic domain in a thin multilayer film, using data in the form of an electron micrograph. Two approaches are illustrated, both of which are based on the concept of two smooth intensity surfaces, one corresponding to the domain and one corresponding to the background, so that the objective is to determine the boundary between the parts of the image where the two surfaces are present. In the first, a hierarchical Bayesian approach, priors are assumed for the set of domain/background states and for the two intensity surfaces, and restoration is carried out using a version of Besag's ICM procedure. It is shown that it is important to initialize the method efficiently. The second approach uses a template-like model for the boundary and also relies on a Bayesian approach. The results from the second approach can be used as an end in themselves or as a way of initializing the more flexible first approach. Several illustrations are provided.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476964

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Machine classification of acoustic waveforms as speech events is often difficult due to context dependencies. Here a vowel recognition task with multiple speakers is studied via the use of a class of modular and hierarchical systems referred to as mixtures-of-experts and hierarchical mixtures-of-experts models. The statistical model underlying the systems is a mixture model in which both the mixture coefficients and the mixture components are generalized linear models. A full Bayesian approach is used as a basis of inference and prediction. Computations are performed using Markov chain Monte Carlo methods. A key benefit of this approach is the ability to obtain a sample from the posterior distribution of any functional of the parameters of the given model. In this way, more information is obtained than can be provided by a point estimate. Also avoided is the need to rely on a normal approximation to the posterior as the basis of inference. This is particularly important in cases where the posterior is skewed or multimodal. Comparisons between a hierarchical mixtures-of-experts model and other pattern classification systems on the vowel recognition task are reported. The results indicate that this model showed good classification performance and also gave the additional benefit of providing for the opportunity to assess the degree of certainty of the model in its classification predictions.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476965

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Several types of reinterview surveys have been designed to measure the different kinds of response errors that occur in survey data. This article is based on the reinterview procedures conducted by the U.S. Census Bureau in its monthly Current Population Survey. We present estimates of the misclassification error rates associated with the original survey data collection procedures and the reinterview methods. A new procedure enables us to estimate the dependence between the error rates of the original survey and the initial nonreconciled reinterview when the reconciled reinterview method is used. Our approach extends to the survey setting methods previously developed for estimating error rates of diagnostic tests. This new approach indicates that the response-variance reinterview methods are effective in replicating the original survey's error rates; however, the reconciled reinterview method creates a substantial dependence between the original and the initial nonreconciled reinterview error rates.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476966

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Numerous statistical studies of the determinants of children's attainments measure the circumstances or events occurring over the childhood period by observations of these variables for a single year or a short duration during childhood. These variables are accepted as proxies for information over the entire childhood period. We explore the reliability of estimated results from studies that use such “window” variables. Because window variables describing intermittent events and discontinuous periods of more persistent characteristics may fail to correspond to variables describing the entire childhood experience, the basic question concerns the extent to which such limited duration information is consistent with that measured over the entire childhood period. We first present an omitted variables model that describes the nature of the “window” problem, and which allows us to measure the consistency of window variables to their longer-duration counterparts. We then use the distinctions revealed by this model to empirically study the potential problems associated with the use of window variables. We use 21 years of data on a sample of 1,705 children from the Michigan Panel Study of Income Dynamics in reduced form models of the determinants of children's schooling and fertility outcomes. We develop four tests of the reliability of estimates using varying window lengths relative to full information on the childhood experience. These include omitted variable likelihood ratio tests, tests of goodness of fit, a sign and significance comparison, and a comparison of the magnitude of the simulated changes using window variables versus those of longer duration.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476967

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Central dimension-reduction subspaces, which characterize the dependence of a response variable on one or more predictors, are developed and then used to guide the construction and interpretation of graphics for regression problems with a binary response variable. Graphical methods requiring neither a link function nor residuals are suggested for both development and criticism of model components implied by the central dimension-reduction subspace.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476968

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

We investigate a natural extension of the profiletplot of Bates and Watts to a general parametric functiong(θ) of the parameters θ in a general maximum likelihood analysis. Although the basic purpose of the extension, called the signed root deviance profile (SRDP), is to construct likelihood ratio (LR) confidence intervals forg(θ), it has various other applications that significantly extend its usefulness. The tangent to the plot of the SRDP at the maximum likelihood estimate ĝ ofg(θ) gives the linear approximation (LA) interval based on the observed information matrix. The plot may be used as a diagnostic tool to compare LA and LR intervals and to suggest transformations ofg(θ) whose LA intervals when inverted are close to the LR intervals forg(θ). The standard way to construct any profile is through repeated optimizations, but problems associated with nonlinear constraints can make this difficult. An alternative method based on integration is presented that avoids these problems. An example and a simulation study are given to illustrate the proposed methods.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476969

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor

摘要:

Instrumental variable estimation in generalized linear measurement error models are studied. For models with canonical link functions, unbiased estimating equations are derived. The maximum likelihood estimator for the normal theory, structural linear instrumental variable model is shown to be a solution to the estimating equations derived herein. Logistic regression is studied in detail. An example is given and a simulation study described for the logistic model based on the Framingham Heart Study data.

ISSN:0162-1459    

DOI:10.1080/01621459.1996.10476970

出版商:Taylor & Francis Group    

年代:1996

数据来源: Taylor