WinBUGS Software Research Articles

Semi-parametric spatial joint modeling of HIV and HSV-2 among women in Kenya.

Several diseases have common risk factors. The joint modeling of disease outcomes within a spatial statistical context may provide more insight on the interaction of diseases both at individual and at regional level. Spatial joint modeling allows for studying of the relationship between diseases and also between regions under study. One major approach for joint spatial modeling is the multivariate conditional autoregressive approach. In this approach, it is assumed that all the covariates in the study have linear effects on the multiple response variables. In this study, we relax this linearity assumption and allow some covariates to have nonlinear effects using the penalized regression splines. This model was used to jointly model the spatial variation of human immunodeficiency virus (HIV) and herpes simplex virus-type 2 (HSV-2) among women in Kenya. The model was applied to HIV and HSV-2 prevalence data among women aged 15–49 years in Kenya, derived from the 2007 Kenya AIDS indicator survey. A full Bayesian approach was used and the models were implemented in WinBUGS software. Both diseases showed significant spatial variation with highest disease burdens occurring around the Lake Victoria region. There was a nonlinear association between age of an individual and HIV and HSV-2 infection. The peak age for HIV was around 30 years while that of HSV-2 was about 40 years. A positive significant spatial correlation between HIV and HSV-2 was observed with a correlation of 0.6831(95% CI: 0.3859, 0.871).

Spatial analysis of Relative Risks for skin cancer morbidity and mortality in Iran, 2008-2010.

One of the most prevalent cancers in whole world is skin cancer and its prevalence is growing. The present research sought to estimate relative risk of morbidity and mortality due to skin cancer. In this cross-sectional study. The required data were gathered from the registered cancer reports of Cancer Control Office in the Center for Non Communicable Disease of the Iranian Ministry of Health (MOH). The data were extracted at province level in the time span of 2008-10. WINBUGS software was used to analyze the data and to identify high risk regions. ArcGIS10 was utilized to map the distribution of skin cancer and to demonstrate high risk provinces by using classic and fully Bayesian models taking into account spatial correlations of adjacent regions separately for men and women. Relative risk of morbidity for women in Yazd and for men in Kurdistan and relative risk of mortality for women in Bushehr and for men in Kohgiluyeh were found to be the highest. Bayesian model due to regarding adjacent regions correlation, have precise estimation in comparing to classical model. More frequent epidemiological studies to enact skin cancer prevention programs. High risk regions in Iran include central and highland regions. Therefore it is suggested that health decision makers enact public education, using anti UV creams and sunglasses for those parts as a short preventing program.

Unobserved Heterogeneity in Hospitality and Tourism Research

Despite the growing complexity of structural equation model (SEM) applications in tourism, it is surprising that most applications have estimated these models without accounting for unobserved heterogeneity. In this article, we aim to discuss the concept of unobserved heterogeneity in more detail, highlighting its serious threats to the validity and reliability of SEMs. We describe a Bayesian finite mixture modeling framework for estimating SEMs while accounting for unobserved heterogeneity. We provide a comprehensive description of this model, and provide guidance on its estimation using the WinBUGS software. We illustrate the importance of unobserved heterogeneity and the finite mixture modeling framework using a didactic application on brand equity where heterogeneity is likely to play an important role because of the differences in how consumers perceive the different dimensions of brand equity. We compare between various models and illustrate the differences between the standard and heterogeneous SEM and discuss the implications for research and practice.

Bayesian optimal design of step stress accelerated degradation testing

This study presents a Bayesian methodology for designing step stress accelerated degradation testing (SSADT) and its application to batteries. First, the simulation-based Bayesian design framework for SSADT is presented. Then, by considering historical data, specific optimal objectives oriented Kullback-Leibler (KL) divergence is established. A numerical example is discussed to illustrate the design approach. It is assumed that the degradation model (or process) follows a drift Brownian motion; the acceleration model follows Arrhenius equation; and the corresponding parameters follow normal and Gamma prior distributions. Using the Markov Chain Monte Carlo (MCMC) method and WinBUGS software, the comparison shows that KL divergence is better than quadratic loss for optimal criteria. Further, the effect of simulation outliers on the optimization plan is analyzed and the preferred surface fitting algorithm is chosen. At the end of the paper, a NASA lithium-ion battery dataset is used as historical information and the KL divergence oriented Bayesian design is compared with maximum likelihood theory oriented locally optimal design. The results show that the proposed method can provide a much better testing plan for this engineering application.

Reducing Monte Carlo error in the Bayesian estimation of risk ratios using log-binomial regression models.

In cohort studies, binary outcomes are very often analyzed by logistic regression. However, it is well known that when the goal is to estimate a risk ratio, the logistic regression is inappropriate if the outcome is common. In these cases, a log-binomial regression model is preferable. On the other hand, the estimation of the regression coefficients of the log-binomial model is difficult owing to the constraints that must be imposed on these coefficients. Bayesian methods allow a straightforward approach for log-binomial regression models and produce smaller mean squared errors in the estimation of risk ratios than the frequentist methods, and the posterior inferences can be obtained using the software WinBUGS. However, Markov chain Monte Carlo methods implemented in WinBUGS can lead to large Monte Carlo errors in the approximations to the posterior inferences because they produce correlated simulations, and the accuracy of the approximations are inversely related to this correlation. To reduce correlation and to improve accuracy, we propose a reparameterization based on a Poisson model and a sampling algorithm coded in R.

TNF-alpha inhibitors for ankylosing spondylitis.

TNF-alpha inhibitors for ankylosing spondylitis.

A Multilevel Higher Order Item Response Theory Model for Measuring Latent Growth in Longitudinal Data.

In educational and psychological testing, individuals are often repeatedly measured to assess the changes in their abilities over time or their latent trait growth. If a test consists of several subtests, the latent traits may have a higher order structure, and traditional item response theory (IRT) models for longitudinal data are no longer applicable. In this study, various multilevel higher order item response theory (ML-HIRT) models for simultaneously measuring growth in the second- and first-order latent traits of dichotomous and polytomous items are proposed. A series of simulations conducted using the WinBUGS software with Markov chain Monte Carlo (MCMC) methods reveal that the parameters could be recovered satisfactorily and that latent trait estimation was reliable across measurement times. The application of the ML-HIRT model to longitudinal data sets is illustrated with two empirical examples.

Relationship Between Hygienic Behavior and Varroa destructor Mites in Colonies Producing Honey or Royal Jelly

Genetic and phenotypic parameters for hygienic behavior, invasion and infestation rates, total and effective reproduction of Varroa destructor in Africanized honeybee colonies producing honey (20 hives) or royal jelly (30 mini-hives) were analyzed. The significance of monthly fixed effects, type of product (honey and royal jelly) and their interactions were verified through generalized linear model procedures. Software WinBugs (Bayesian Inference Using Gibbs Sampling) with Bayesian inference was employed for (co)variance estimates. The average values for colonies producing honey or royal jelly were 74.38 and 71.40% for hygienic behavior in 24 hours; infestation rates 8.30 and 11.40%; invasion rates 9.50 and 7.50%; total reproduction 1.02 and 0.55%; effective reproduction was 0.62 and 0.33%, respectively. The additive genetic variance for invasion (0.16), total reproduction (0.25) and effective reproduction (0.94) rates of the mite were higher than estimates for hygienic behavior in 24 hours (0.05) and infestation rate (0.04). Mean heritability for hygienic behavior, infestation and invasion rates, total and effective reproduction of the mite were .58, 0.54, 0.56, 0.63 and 0.61, respectively. The genetic correlation of -0.48 for hygienic behavior with total reproduction rate of Varroa destructor shows that hygienic behavior may be the most interesting trait for selection. Besides a heritability of high magnitude, when combined with the total reproduction rate of the mite, it has a high and antagonistic correlation. Consequently, in cases of high infestation of Varroa destructor, the selection for hygienic behavior would decrease the reproduction rate of the mite.

Joint longitudinal and survival data modelling: an application in anti-diabetes drug therapeutic effect

The longitudinal and survival analyses are useful tools in the exploration of drug trial data. In both cases the challenge is to deal with correlated repeated observations. Here, the joint modelling for longitudinal and survival data has been carried out via Markov chain Monte Carlo (MCMC) method in type 2 diabetes clinical trials to compare different combinations of drugs, viz. Metformin plus Pioglitazone and Gliclazide plus Pioglitazone. Despite the complexity of the model it has been found relatively easier to implement with WinBugs software. The results have been computed and compared with software R. In both types of the analyses it has been found that no estimates of treatment appear to have significant effect on the evolution of the matter of HBA1c, neither on the longitudinal part nor on the survival one. The Bayesian approach has been considered as an extended tool with classical approach for estimation of clinical trial data analysis.

A mixed treatment comparison of gabapentin enacarbil, pramipexole, ropinirole and rotigotine in moderate-to-severe restless legs syndrome

Objective:A mixed treatment comparison (MTC) was performed to investigate the relative efficacy and safety of licensed pharmaceuticals for moderate-to-severe restless legs syndrome (RLS).Methods:RLS trials published over the past 10 years were identified via systematic literature searches of MEDLINE, Embase, Cochrane CENTRAL, and manufacturers’ websites. MTC was performed with WinBUGS software using a Bayesian approach. Identified primary outcomes: change in International RLS Study Group Rating Scale (IRLS) at week 12 and end of maintenance (EoM). Secondary outcomes: IRLS and Clinical Global Impression – Improvement Scale (CGI-I) responders, RLS-6 items and adverse events (AEs).Results:Twenty-eight clinical trials were identified. Fifteen were included in the primary analysis. Indirect comparisons were established among gabapentin enacarbil, pramipexole, ropinirole, rotigotine and placebo. Overall, the four active treatments showed similar efficacies as assessed by changes in IRLS scores, IRLS responders, CGI-I responders, and RLS-6 scores. The sole exception was change in IRLS at week 12, for which rotigotine was likely more efficacious than ropinirole (mean difference: −2.52 [95% CrI: −4.74, −0.40]). Indirect comparisons on safety endpoints indicated ropinirole was associated with a higher risk of nausea than the other agents, and was more likely to result in discontinuations due to lack of efficacy than pramipexole. Nausea was likely more frequent with pramipexole than gabapentin enacarbil, and rotigotine was more likely to result in discontinuation due to AEs than ropinirole and pramipexole.Conclusions:This MTC confirmed the superiority of gabapentin enacarbil, pramipexole, ropinirole, and rotigotine above placebo in alleviating RLS symptoms. Compared to ropinirole, rotigotine showed some additional benefit in terms of change in IRLS at Week 12. Choice of RLS drugs requires careful evaluation of effectiveness and safety profiles in clinical practice. Due to lack of head-to-head trials, inconsistency could not be assessed in our analysis. Head-to-head trials on a more homogeneous population are needed to validate the MTC results.

Modeling Electricity Price Using A Threshold Conditional Autoregressive Geometric Process Jump Model

Electricity market prices are highly volatile and often have high spikes. Both government authorities and market participants require sophisticated models and techniques for forecasting future prices and managing relevant financial risks in such a volatile market. This article extends the conditional autoregressive geometric process (CARGP) model (Chan et al., 2012) to the CARGP model with thresholds and jumps, which is abbreviated as CARGP-TJ model in this article. We will demonstrate that the proposed CARGP-TJ model not only captures the unique features of the electricity price but also performs better than other existing models. For robustness consideration, a heavy-tailed error distribution is adopted. Model implementation relies on the powerful Bayesian Markov chain Monte Carlo simulation techniques via WinBUGS software. The analysis of the daily maximum electricity prices of the New South Wales, Australia reveals that the proposed CARGP-TJ model captures the price spikes well for both in-sample estimation and out-of-sample forecast.

A Bayesian Analysis of Unobserved Heterogeneity for Unemployment Duration Data in the Presence of Interval Censoring

In this paper, we discuss Bayesian inference of unobserved heterogeneity for unemployment duration data in the presence of right and interval-censoring, and non-proportionality. We employ accelerated failure time models with three different distributional assumptions: log-logistic, log-normal, and Weibull models, and use members of an exponential family of distributions for considering unobserved heterogeneity. We adopt a Bayesian approach, using Markov Chain Monte Carlo via WinBUGS software, to analyze the data. The proposed approach is illustrated using the unemployment duration data set of Iran in 2009. A sensitivity analysis using different latent variable models of the exponential family is also considered. After checking convergence, using the Gelman-Rubin diagnostic test, we compared different distributional assumptions using the DIC3 criterion. Our findings reveal significant discrepancies in unemployment duration based on different covariates for the sample population of Iran in 2009.

Multivariate generalized Poisson geometric process model with scale mixtures of normal distributions

Abstract This paper proposes a new model named as the multivariate generalized Poisson log- t geometric process (MGPLTGP) model to study multivariate time-series of counts with overdispersion or underdispersion, non-monotone trends within each time-series and positive or negative correlation between pairs of time-series. This model assumes that the multivariate counts follow independent generalized Poisson distributions with an additional parameter to adjust for different degrees of dispersion including overdispersion and underdispersion. Their means after discounting the trend effect geometrically by ratio functions form latent stochastic processes and follow a multivariate log- t distribution with a flexible correlation structure to capture both positive correlation and negative correlation. By expressing the multivariate Student’s t -distribution in scale mixtures of normals, the model can be implemented through Markov chain Monte Carlo algorithms via the user-friendly WinBUGS software. The applicability of the MGPLTGP model is illustrated through an analysis of the possession and/or use of two illicit drugs, amphetamines and narcotics in New South Wales, Australia.

A Poisson geometric process approach for predicting drop-out and committed first-time blood donors

A Poisson geometric process (PGP) model is proposed to study individual blood donation patterns for a blood donor retention program. Extended from the geometric process (GP) model of Lam [16], the PGP model captures the rather pronounced trend patterns across clusters of donors via the ratio parameters in a mixture setting. Within the state-space modeling framework, it allows for overdispersion by equating the mean of the Poisson data distribution to a latent GP. Alternatively, by simply setting, the mean of the Poisson distribution to be the mean of a GP, it has equidispersion. With the group-specific mean and ratio functions, the mixture PGP model facilitates classification of donors into committed, drop-out and one-time groups. Based on only two years of observations, the PGP model nicely predicts donors’ future donations to foster timely recruitment decision. The model is implemented using a Bayesian approach via the user-friendly software WinBUGS.

An application of MCMC simulation in mortality projection for populations with limited data

IIn this paper, we investigate the use of Bayesian modeling and Markov chain Monte Carlo (MCMC) simulation, via the software WinBUGS, to project future mortality for populations with limited data. In particular, we adapt some extensions of the Lee-Carter m

Moderated Mediation Model Analysis Based on Asymmetric Interval

Integrating moderation and mediation models has been a popular topic of study in social science methodology. A moderated mediation model is a combination of both moderation and mediation models. When a mediation effect is moderated by a moderator, the effect is termed moderated mediation and the model is a moderated mediation model. There are three common types of moderated mediation models: first-stage moderated mediation, second-stage moderated mediation, and dual-stage moderated mediation. Three methods of testing moderated mediation models are based on the Bootstrap asymmetric confidence interval and the Bayesian asymmetric credible interval. These methods include a subgroup approach, difference approach, and products of coefficients approach. The products of coefficients approach is recommended to analyze first-stage or second-stage moderated mediation models. The difference approach is recommended to analyze dual-stage moderated mediation models. A simulation study found that the Bayesian credible interval without informative prior yielded higher power than the Bootstrap percentile confidence interval, and comparable power to the Bootstrap bias-corrected confidence interval. In the present study, we used an example to illustrate how to conduct the proposed procedure by using MPLUS and WinBUGS software. MPLUS and WinBUGS programs are provided in the appendix to facilitate the implementation of the Bootstrap method to analyze moderated mediation effects. The programs can be managed easily by empirical researchers. To explore the present situation and problems of using three moderated mediation analysis methods in psychological researches in China, this article reviewed 12 studies published in Chinese psychological journals, retrieved from the CNKI database, since 2009 to 2014.

Multiple treatment and indirect treatment comparisons: An overview of network meta-analysis.

Randomized control trials and its meta-analysis has occupied the pinnacle in levels of evidence available for research. However, there were several limitations of these trials. Network meta-analysis (NMA) is a recent tool for evidence-based medicine that draws strength from direct and indirect evidence generated from randomized control trials. It facilitates comparisons across multiple treatment options, direct comparisons of which have not been attempted till date due to multitude of reasons. These indirect treatment comparisons of randomized controlled trials are based on similarity and consistency assumptions that follow Bayesian or frequentist statistics. Most NMAuntil date use Microsoft Windows WinBUGs Software for analysis which relies on Bayesian statistics. Methodology of NMA is expected to undergo further refinements and become robust with usage. Power and precision of indirect comparisons in NMA is a concern as it is dependent on effective number of trials, sample size and complete statistical information. However, NMA can synthesize results of considerable relevance to experts and policy makers.

Variation in Specificity of HIV Rapid Diagnostic Tests over Place and Time: An Analysis of Discordancy Data Using a Bayesian Approach

BackgroundRecent trends to earlier access to anti-retroviral treatment underline the importance of accurate HIV diagnosis. The WHO HIV testing strategy recommends the use of two or three rapid diagnostic tests (RDTs) combined in an algorithm and assume a population is serologically stable over time. Yet RDTs are prone to cross reactivity which can lead to false positive or discordant results. This paper uses discordancy data from Médecins Sans Frontières (MSF) programmes to test the hypothesis that the specificity of RDTs change over place and time. MethodsData was drawn from all MSF test centres in 2007-8 using a parallel testing algorithm. A Bayesian approach was used to derive estimates of disease prevalence, and of test sensitivity and specificity using the software WinBUGS. A comparison of models with different levels of complexity was performed to assess the evidence for changes in test characteristics by location and over time. Results106, 035 individuals were included from 51 centres in 10 countries using 7 different RDTs. Discordancy patterns were found to vary by location and time. Model fit statistics confirmed this, with improved fit to the data when test specificity and sensitivity were allowed to vary by centre and over time. Two examples show evidence of variation in specificity between different testing locations within a single country. Finally, within a single test centre, variation in specificity was seen over time with one test becoming more specific and the other less specific.ConclusionThis analysis demonstrates the variable specificity of multiple HIV RDTs over geographic location and time. This variability suggests that cross reactivity is occurring and indicates a higher than previously appreciated risk of false positive HIV results using the current WHO testing guidelines. Given the significant consequences of false HIV diagnosis, we suggest that current testing and evaluation strategies be reviewed.

Using WinBUGS to Study Family Frailty in Child Mortality, with an Application to Child Survival in Ivory Coast

This article analyzes the effects of unobserved family heterogeneity in children survival times through a Bayesian approach. We rely on survey data from Ivory Coast and use a proportional hazard model with multiplicative random effect. With such a model, the usual assumption of independence of observations is avoided. The posterior distributions of the parameters are estimated through a Gibbs sampler algorithm using the WinBUGS software. This technique overcomes the possible local convergence problem observed with the commonly used Expectation-Maximization method.

A caution about using deviance information criterion while modeling traffic crashes

The Poisson-Gamma (PG) or negative binomial (NB) model still remains the most popular method used for analyzing count data. In the software WinBUGS (or any other software used for Bayesian analyses), there are different ways to parameterize the NB model. In general, either a PG (based on the Poisson-mixture) or a NB (based on the Pascal distribution) modeling framework can be used to relate traffic crashes to the explanatory variables. However, it is important to note that the way the model is parameterized will influence the output of the Deviance Information Criterion (DIC) values. The objective of this short study is to document the difference between the PG and NB models in the estimation of the DIC. This is especially important given that the NB/PG model is still the most frequently used model in highway safety research and applications. To accomplish the study objective, PG and NB models were developed using the crash data collected at 4-legged signalized intersections in Toronto, Ont. The study results showed that there is a considerable difference in the estimation of the DIC values between the two models. It is thus recommended not to consider the DIC as the sole model selection criterion and the comparison should be done only between the models that have similar parameterization. Other alternatives such as Bayes Factors, Posterior predictive performance criterion, Bayesian Information Criterion (BIC), among others need to be considered in addition to the DIC in the model selection.