摘要:

In this paper we analyze serial CD4 T-cell measurements from the Los Angeles portion of the Multicenter AIDS Cohort Study. Our emphasis is on developing a plausible and parsimonious model to describe the stochastic process underlying the patterns of CD4 measurements. The stochastic process that we use enables us to investigate the concept of derivative tracking, for which it is assumed that the rank order of the individual's slopes is maintained over time. A general model for the analysis of longitudinal repeated measures data iswhereYi(tij) is the measurement of subjectiat timetij, X(tij)α represents fixed effect terms,Z(tij)birepresents random effect terms,Wi(tij) is a stochastic process allowing correlation between measurements, and εijis measurement error. In the simplest case,X(tij) andZ(tij) contain the times of measurements. ForWi(tij), we use a two-parameter integrated Ornstein-Uhlenbeck (OU) process. The OU process is the continuous mean zero Gaussian Markov process, which includes Brownian motion and white noise as special limiting cases. This model is a continuous-time version of an AR(1) process for the deviations of the derivative ofyfrom the expected derivative ofywith respect tot.This approach is flexible and tractable as the covariance structure has a closed-form expression. The model allows unequally spaced observations and can be generalized to multivariate responses. This model enables one to assess whether individuals maintain their trajectories; that is, whether their slope ofYtracks. We find no evidence in the data that the slopes of the CD4 values track.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476806

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

In the AIDS Clinical Trial Group randomized trial 002 comparing the effect of high-dose versus low-dose 3-azido-3-deoxythymidine (AZT) on the survival of acquired immunodeficiency syndrome (AIDS) patients, the median survival in the low-dose arm exceeded that in the high-dose arm. But subjects in the low-dose AZT arm received significantly more prophylaxis therapy for pneumocystis carinii pneumonia (PCP), a nonrandomized treatment, than those in the high-dose AZT arm. Thus the improved median survival in the low-dose arm might represent either the benefits associated with avoiding the toxicity of high-dose AZT therapy or the benefits of receiving prophylaxis therapy. The authors use structural nested failure time (SNFT) models to estimate the survival curves that would have been observed if the PCP prophylaxis experience in the high-dose and low-dose treatment arms had been similar. Our simplest models relate a subject's observed time of death and observed prophylaxis therapy to the time that the subject would have died if prophylaxis therapy had been withheld. Conditional on certain assumptions and our model, we infer that survival in the low-dose arm would have still exceeded that in the high-dose arm even if the two arms had been given identical amounts of prophylaxis. Under the same assumptions, we also find the data are consistent with continuous prophylaxis therapy increasing survival by 16% or decreasing survival by 18% at the 95% confidence level (p= .85).

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476807

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

The National Center for Health Statistics makes available data from three national health evaluation surveys that it has conducted since 1960: NHES I (1960-1962), NHANES I (1971-1975), and NHANES II (1976-1980). There has been considerable interest in using these data to assess secular trends in cardiovascular risk factors such as blood pressure (BP). Unfortunately, underlying trends in BP are confounded with trends in physician treatment of hypertension over the same period; in the early 1960s it was rare to treat hypertension, whereas by the late 1970s it had become quite common. Our approach to estimating the underlying trends is to take untreated BP to be the variable of interest and to consider it missing in those subjects who are under treatment. We then use a multiple-imputation scheme to construct estimates of trend parameters that adjust for the incompleteness of the original data. Because our imputations depend on certain model features that the data cannot address, we form estimates under different models and compare the results. Our analyses suggest that trend estimates are sensitive to the assumed model, and naive estimates that do not adjust for treatment trends appear to be overly optimistic.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476808

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

We use Oregon state data (1981-1990) on medical malpractice claims to develop a random-effects probit model for vulnerability to a medical malpractice claim in practice yeark(k= 1, 2, …,ni) for physiciani(i= 1, 2, …,Nphysicians in the sample) conditional on anni×pcovariate matrixWithat contains a mixture ofptime-varying and time-invariant covariates. In this application, time-invariant covariates were physician sex and specialty (surgical versus nonsurgical). Time-varying covariates were age, the cumulative amount of risk management education (i.e., number of courses) taken by physicianito yeark, and prior claim history. In addition, the model incorporates a random effect of “claim vulnerability” assumed to be normally distributed in the population of physicians. This random effect represents unobservable and/or unmeasured characteristics that place one physician at greater risk for experiencing a medical malpractice claim than another physician. In addition, we also determine if the effects of risk management training on claim vulnerability differ before and after the physician's first malpractice claim. Results of the analysis reveal that (1) there is a sizable random physician effect; (2) risk increases between age 40 to 60; (3) physicians in a surgical specialty are at increased risk; (4) male physicians are at greater risk than female physicians; (5) risk increases following an initial claim, particularly in the year subsequent to the initial claim, and (6) some beneficial effects of risk management education are observed in physicians with a prior claim history, particularly those in anesthesiology and obstetrics and gynecology.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476809

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

Adaptive clinical trials have attracted the attention of statisticians because they allow for information accrued early in the trial to influence the allocation of treatment of later enrolled patients, thus allowing those patients an increased likelihood of receiving the better treatment. Few adaptive clinical trials have been reported in the literature; the purpose of this article is to describe the rationale, design, and analysis of a recently completed adaptive trial. Simulations to address sample size issues are presented. A Bayesian and a frequentist randomization analysis are discussed. The randomization analysis addresses the use of an adaptive allocation scheme, a delayed response, and a surrogate response. In this trial, the differences between the Bayesian and randomization inference were small. Some suggestions for future implementation of adaptive clinical trials and future research areas are provided.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476810

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

This article checks, using both real and simulated data, the effectiveness of modern adaptive techniques to track the parameters of time-varying dynamic models. The real case studies concern a bone marrow transplant data set published by Tong, the gas furnace model of Box and Jenkins, and two series of West German interest rates. Simulation studies focus on ARX models with smoothly and suddenly changing parameters. The general approach is to compare the fitting-forecasting performance of classical and adaptive methods, holding fixed the order of the models. At the methodological level, the basic step is taken by unifying known estimators, such as recursive least squares and Kalman filter, into a general algorithm. Next, the problem of optimal design of the tracking coefficients (such as discounting factors and learning rates), is solved by optimizing a quadratic functional based on one-step-ahead prediction errors. All applications show that adaptive modeling, based on the design and the optimization of recursive algorithms, leads to significant improvements of the forecasting performance.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476811

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

In atomic and molecular spectroscopy, measured transition energies represent differences between the energy levels of the emitting or absorbing system. In complex spectra of such transitions, the assignment of the pair of states corresponding to an observed transition energy is a difficult process. These assignments evolve in stages. At each stage, some assignments will be erroneous. Robust fitting methods have proven extremely helpful for finding gross errors and for obtaining reliable estimates of energy levels in the analysis of molecular tritium (T2) spectra. The robust results can be used successfully to obtain new assignments of transition energies to energy levels. In statistical terms, a highly unbalanced two-way analysis of variance model results. Some specific robustness problems that call for further research are mentioned. An iterative algorithm that is fast and avoids storage problems, even for large data sets, is described. This algorithm can be used for both least squares and robust fitting of incomplete two-way tables and similar designs.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476812

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476813

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

In recent years, the focus of study in smoothing parameter selection for kernel density estimation has been on the univariate case, while multivariate kernel density estimation has been largely neglected. In part, this may be due to the perception that calibrating multivariate densities is substantially more difficult. In this article, we explicitly derive and compare multivariate versions of the bootstrap method of Taylor, the least-squares cross-validation method developed by Bowman and Rudemo, and a biased cross-validation method similar to that of Scott and Terrell for multivariate kernel estimation using the product kernel estimator. The theoretical behavior of these cross-validation algorithms is shown to improve (surprisingly) as the dimension increases, approaching the best rate ofO(n−1/2). Simulation studies suggest that the new biased cross-validation method performs quite well and with reasonable variability as compared to the other two methods. Bivariate examples with heart disease and ozone data are given to illustrate the behavior of these algorithms.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476814

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor

摘要:

Markov chain sampling schemes generate dependent observations {Θi, 0 ≤ i ≤ n} from a full joint posterior distribution π(θdata). Frequently, only certain marginals of this full posterior density are of interest; thus an interesting problem is how to estimate the marginal posterior densities based on the dependent observations {Θi, 0 ≤ i ≤ n} from π(θ data). We propose a new importance-weighted marginal density estimation (IWMDE) method. An IWMDE is obtained by averaging many dependent observations of the ratio of the full joint posterior densities multiplied by a weighting conditional densityw.The asymptotic properties for the IWMDE and the guidelines for choosing a weighting conditional densityware also considered. A bivariate normal model and a constrained linear multiple regression model are used to illustrate how to derive the IWMDE's for the marginal posterior densities.

ISSN:0162-1459    

DOI:10.1080/01621459.1994.10476815

出版商:Taylor & Francis Group    

年代:1994

数据来源: Taylor