Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Bayesian modelling”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 883 records · Page 49Linked to original sources
PubMed2004

Hierarchical models for tumor xenograft experiments in drug development.

In cancer drug development, demonstrated anticancer activity in animal models is an important step to bring a promising compound to clinic. Proper design and analysis of experiments using laboratory animals have received increasing attention recently. These experiments involve informatively censored longitudinal data with small samples. The problem is further complicated because of order constraints due to the intrinsic growth of control tumors without treatment. This article proposes a Bayesian hierarchical model to analyze informatively censored longitudinal data while accounting for the parameter constraints and providing valid small sample inference. We adopt a noniterative sampling approach, the inverse Bayes formulae (IBF) sampler, to generate independent posterior samples, which avoids convergence problems associated with Markov chain Monte-Carlo methods. To effectively deal with the restricted parameter problem, we use a linear transformation to simplify the constraints and exploit the IBF method to generate random samples from truncated multivariate normal distributions. Because diffuse priors are used, the posterior modes approximate the maximum likelihood estimates well, and the hierarchical model can be considered as an extended mixed-effects model. A real xenograft experiment on a new treatment is analyzed by using the proposed method.

Algorithms↗
PubMed1993

Can pharmacokinetic dosing decrease nephrotoxicity associated with aminoglycoside therapy.

A randomized, controlled clinical trial was performed to determine whether individualized dosing by use of Bayesian pharmacokinetic modeling could decrease nephrotoxicity accosted with aminoglycoside therapy. Two hundred forty-three patients receiving aminoglycosides for suspected or proven infection were randomly assigned to one of three groups: usual physician-directed dosing (Group 1), pharmacist-assisted dosing (Group 2), or pharmacist-directed dosing (Group 3). Dosing in Groups 2 and 3 was based on a Bayesian pharmacokinetic dosing program, whereas Group 1 served as the control group. Individualized dosing resulted in higher mean postinfusion (peak) serum aminoglycoside levels, higher ratios of mean peak level to minimum inhibitory concentration (peak/MIC ratios), and a trend toward lower trough serum levels. Milligrams per dose were higher and number of doses per day was lower in the pharmacist-dosed groups. However, the incidence of nephrotoxicity (> or = 100% increase in serum creatinine) was not different among the three groups (16, 27, and 16% in Groups 1, 2, and 3, respectively). Similarly, severity of toxicity was not affected by the dosing intervention. Risk factors for toxicity included duration of therapy, shock, treatment with furosemide, older age, and liver disease. After controlling for these factors, the dosing intervention still had no effect on nephrotoxicity. It was concluded that Bayesian pharmacokinetic dosing did not decrease the risk of nephrotoxicity associated with aminoglycoside therapy.

Aged↗
PubMed2001

Population pharmacokinetic modelling of carbamazepine by using the iterative Bayesian (IT2B) and the nonparametric EM (NPEM) algorithms: implications for dosage.

OBJECTIVE: To estimate individual and population postinduction pharmacokinetics of carbamazepine (CBZ) in epileptic adult and paediatric patients who received chronic CBZ monotherapy. METHODS: We have used the USC*PACK collection of PC programs for the estimations. The preinduction CBZ metabolism was also estimated in 16 volunteers after a single dose of CBZ (200 mg). We used a linear one-compartmental model with oral absorption and found the pharmacokinetic parameter values of CBZ behaviour to be in good agreement with those reported earlier. RESULTS: Serum CBZ concentrations correlated poorly with daily doses in both the adult and child populations. Because of the diversity within the population, use of the mean population model without knowledge of an individual patient's pharmacokinetic characteristics gives poor prediction. In contrast, the individual Bayesian posterior models gave good prediction for all subjects in the population, due to the removal of the interindividual variability. CONCLUSION: This approach permits one to individualize drug therapy for patients even when only sparse therapeutic drug monitoring (TDM) data are available. Future individual CBZ serum level predictions were acceptable from a clinical point of view (mean absolute error = 13.2 +/- 9.7%). The optimal sampling strategy approach helped to design an optimal cost-effective TDM protocol for CBZ therapy management.

Adult↗
PubMed2025

Generative model for the first cell fate bifurcation in mammalian development.

The first cell fate bifurcation in mammalian development directs cells toward either the trophectoderm (TE) or inner cell mass (ICM) compartments in pre-implantation embryos. This decision is regulated by the subcellular localization of a transcriptional co-activator YAP and takes place over several progressively asynchronous cleavage divisions. As a result of this asynchrony and variable arrangement of blastomeres, reconstructing the dynamics of the TE/ICM cell specification from fixed embryos is extremely challenging. To address this, we developed a live-imaging approach and applied it to measure pairwise dynamics of nuclear YAP and its direct target genes, CDX2 and SOX2, which are key transcription factors of the TE and ICM, respectively. Using these datasets, we constructed a generative model of the first cell fate bifurcation, which reveals the time-dependent statistics of the TE and ICM cell allocation. In addition to making testable predictions for the joint dynamics of the full YAP/CDX2/SOX2 motif, the model revealed the stochastic nature of the induction timing of the key cell fate determinants and identified the features of YAP dynamics that are necessary or sufficient for this induction. Notably, temporal heterogeneity was particularly prominent for SOX2 expression among ICM cells. As heterogeneities within the ICM have been linked to the initiation of the second cell fate decision in the embryo, understanding the origins of this variability is of key significance. The presented approach reveals the dynamics of the first cell fate choice and lays the groundwork for dissecting the next cell fate decisions in mouse development.

Animals↗
PubMed1991

Estimation of relative potency with sequential dilution errors in radioimmunoassay.

Sequential dilution is a very common procedure in radioimmunoassay, in which the dilution error will be accumulated from the highest to the lowest concentration. A simulated example in relative potency determination is used to demonstrate the potentially wrong conclusion that can be drawn, when the dilution error is not properly included in the model. A Bayesian method is used and an alternative approximation via maximum likelihood is proposed. An alternative experimental design is recommended to increase the precision of the inference.

Bayes Theorem↗
PubMed2006

Optimal selection of individuals for BRCA mutation testing: a comparison of available methods.

PURPOSE: Several methods have been described that estimate the likelihood that a family history of cancer is a result of a mutation in the BRCA1 or BRCA2 genes. We examined the performance of six different methods with the aim of identifying an optimal strategy for selecting individuals for mutation testing in clinical practice. PATIENTS AND METHODS: Two hundred fifty-seven families who had completed BRCA1 and BRCA2 mutation screening were assessed by six models representing the major methodologies used to assess the likelihood of a pathogenic mutation. The performance of each method as a selection criterion was compared with the result of mutation testing to produce sensitivity, specificity, and receiver operating curve data. The impact of incorporating breast cancer pathology data in the assessment was also analyzed. RESULTS: The highest accuracy was achieved by the Bayesian probabilistic model (BRCAPRO). The formal probabilistic methods were significantly more accurate than clinical scoring methods. The methods were further improved by the incorporation of information on breast cancer pathology (tumor grade and estrogen receptor/progesterone receptor status). The resulting combined probability figure was highly accurate when selecting individuals for BRCA1 testing. Some BRCA2 mutation carriers were missed by all of the models examined. CONCLUSION: Formal probabilistic models provide significantly greater accuracy in the selection of families for gene testing than the use of clinical criteria or scoring methods. The accuracy is further enhanced by incorporating information on the pathology of breast cancers occurring in the families.

Adult↗
PubMed2003

Approximate cross-validatory predictive checks in disease mapping models.

When fitting complex hierarchical disease mapping models, it can be important to identify regions that diverge from the assumed model. Since full leave-one-out cross-validatory assessment is extremely time-consuming when using Markov chain Monte Carlo (MCMC) estimation methods, Stern and Cressie consider an importance sampling approximation. We show that this can be improved upon through replication of both random effects and data. Our approach is simple to apply, entirely generic, and may aid the criticism of any Bayesian hierarchical model.

Bayes Theorem↗
PubMed2000

Modeling markers of disease progression by a hidden Markov process: application to characterizing CD4 cell decline.

Multistate models have been increasingly used to model natural history of many diseases as well as to characterize the follow-up of patients under varied clinical protocols. This modeling allows describing disease evolution, estimating the transition rates, and evaluating the therapy effects on progression. In many cases, the staging is defined on the basis of a discretization of the values of continuous markers (CD4 cell count for HIV application) that are subject to great variability due mainly to short time-scale noise (intraindividual variability) and measurement errors. This led us to formulate a Bayesian hierarchical model where, at a first level, a disease process (Markov model on the true states, which are unobserved) is introduced and, at a second level, the measurement process making the link between the true states and the observed marker values is modeled. This hierarchical formulation allows joint estimation of the parameters of both processes. Estimation of the quantities of interest is performed via stochastic algorithms of the family of Markov chain Monte Carlo methods. The flexibility of this approach is illustrated by analyzing the CD4 data on HIV patients of the Concorde clinical trial.

Algorithms↗
PubMed1993

A probabilistic rule-based expert system.

This paper explores a medical expert system combining techniques of Bayesian network modelling with ideas of weighted inference rules. The weights of the individual rules can be estimated objectively from a training set of actual cases; and they can be used in a Monte Carlo stimulation to estimate objectively conditional probabilities of diagnosis given particular combinations of symptoms. The paper describes and evaluates a medical expert system built according to this design. The diagnostic accuracy of the program was found to be similar to that obtained through the usual application of Bayes theorem with the assumption of conditional independence of symptoms given disease, even though the Bayesian classifier has more than 70 times as many numerical parameters. The method may be promising in cases where small training sets do not permit accurate estimation of large numbers of parameters.

Abdominal Pain↗
PubMed2005

The ProDom database of protein domain families: more emphasis on 3D.

ProDom is a comprehensive database of protein domain families generated from the global comparison of all available protein sequences. Recent improvements include the use of three-dimensional (3D) information from the SCOP database; a completely redesigned web interface (http://www.toulouse.inra.fr/prodom.html); visualization of ProDom domains on 3D structures; coupling of ProDom analysis with the Geno3D homology modelling server; Bayesian inference of evolutionary scenarios for ProDom families. In addition, we have developed ProDom-SG, a ProDom-based server dedicated to the selection of candidate proteins for structural genomics.

Computer Graphics↗
PubMed2001

Bayesian estimation and test for factor analysis model with continuous and polytomous data in several populations.

The main purpose of this paper is to develop a Bayesian approach for the multisample factor analysis model with continuous and polytomous variables. Joint Bayesian estimates of the thresholds, the factor scores and the structural parameters subjected to some simple constraints across groups are obtained simultaneously. The Gibbs sampler is used to produce the joint Bayesian estimates. It is shown that the conditional distributions involved in the implementation are the familiar uniform, gamma, normal, univariate truncated normal and Wishart distributions. The Bayes factor is introduced to test hypotheses involving constraints among the structural parameters of the factor analysis models across groups. Two procedures for computing the test statistics are developed, one based on the Schwarz criterion (or Bayesian information criterion), while the other computes the posterior densities and likelihood ratios by means of draws from the appropriate conditional distributions via the Gibbs sampler. The empirical performance of the proposed Bayesian procedure and its sensitivity to prior distributions are illustrated by some simulation results and two real-life examples.

Bayes Theorem↗
PubMed2005

Inferring gene transcriptional modulatory relations: a genetical genomics approach.

Bayesian network modeling is a promising approach to define and evaluate gene expression circuits in diverse tissues and cell types under different experimental conditions. The power and practicality of this approach can be improved by restricting the number of potential interactions among genes and by defining causal relations before evaluating posterior probabilities for billions of networks. A newly developed genetical genomics method that combines transcriptome profiling with complex trait analysis now provides strong constraints on network architecture. This method detects those chromosomal intervals responsible for differences in mRNA expression using quantitative trait locus (QTL) mapping. We have developed an efficient Bayesian approach that exploits the genetical genomics method to focus computational effort on the most plausible gene modulatory networks. We exploit a dense marker map for a genetic reference population (GRP) that consists of 32 BXD strains of mice made by intercrossing two progenitor strains--C57BL/6J and DBA/2J. These progenitors differ at approximately 1.3 million known single nucleotide polymorphisms (SNPs), all of which can be exploited to estimate the probability that a gene contains functional polymorphisms that segregate within the GRP. We constructed 66 candidate networks that include all the candidate modulator genes located in the 209 statistically significant trans-acting QTL regions. SNPs that distinguish between the two progenitor strains were used to further winnow the list of candidate modulators. Bayesian network was then used to identify the genetic modulatory relations that best explain the microarray data.

Algorithms↗
PubMed1994

CPT-11: population pharmacokinetic model and estimation of pharmacokinetics using the Bayesian method in patients with lung cancer.

In this study, we aimed to develop a population pharmacokinetic model for CPT-11 and to use the Bayesian method to estimate CPT-11 pharmacokinetic parameters in each of 43 patients who received combined therapy consisting of CPT-11 and etoposide. The group was divided into first and second data sets of 30 and 13 patients, respectively. We developed a population pharmacokinetic model of CPT-11 based on the first data set. The individual pharmacokinetic parameters [area under the concentration curve (AUC) and clearance (CL)] were subsequently estimated by using the Bayesian method on the second data set. Plasma CPT-11 concentrations were measured by high-performance liquid chromatography, and compartmental pharmacokinetic models were fitted by the Bayesian method. The population pharmacokinetic model was developed by using the nonlinear mixed effect model. We selected the volume of the central compartment (Vc), CL, and distribution rate constants (K12, K21) as population pharmacokinetic parameters. The population mean values (CV%) of Vc, CL, K12, and K21 were, respectively, 31.8 (15.7%) liter/m2, 14.1 (27.8%) liter/h/m2, 1.1 (8.4%)/h, and 0.41 (30.3%)/h. Residual intraindividual variability was 22.9%. The optimal sampling regime for estimation of the AUC and CL in using the Bayesian method was the two time points of 1 and 8 h post infusion. The mean predictive error, the mean absolute predictive error, and the root mean squared error were -3.3, 9.4, 3.2% (AUC) and 6.3, 10.0, 3.5% (CL), respectively. We concluded that the AUC and CL of CPT-11 could be estimated from plasma concentrations at two times by using the Bayesian method.

Bayes Theorem↗
PubMed2006

Statistical methods for geographical surveillance in veterinary epidemiology.

Spatial clustering and cluster detection are statistical analysis developed to address relevant scientific hypothesis. The difficulty stays in the large number of alternative hypothesis due to the different mechanisms that could generate the anomalous cases aggregation. We review methods for marked point data (case/control) aimed to describe spatial intensity of disease risk, to test for randomness and to locate significant excesses. Bayesian Gaussian Spatial Exponential models are used to illustrate probabilistic aspects and the link with simpler non parametric tools are shown. We develop an informal guideline to the analysis and used data on faecal contamination and dog parasitic diseases in the city of Naples, Italy. Kernel density estimation resulted very sensitive to bandwidth choice and overemphasized localized excess, Ripley'K function and Cuzick-Edwards test were very consistent each other while the SatScan failed to detect excesses. The spatial range was around 600 meters and justifies several small clusters. Bayesian models were very powerful in reconstructing the phenomenon and allow inference on model parameters in good agreement with the non parametric analysis.

Algorithms↗
PubMed1995

Meta-analysis: weighing the evidence.

Use of meta-analytical (quantitative overview) techniques is now commonplace in a wide range of medical research contexts, with a rapid rise in its frequency of use being particularly apparent in the last decade. The history of meta-analyses in other fields, particularly psychology and educational research, is longer. Many methods have been proposed and used, from crude 'vote counting' of studies showing significant or non-significant results, through methods for combination of effect size estimates based on fixed or random effects models, to general linear mixed models and Bayesian methods. The history of meta-analysis and the advantages and disadvantages of various approaches to it are briefly reviewed in this paper, with reference to pharmaceutical product licence applications, other reviews of clinical trials and epidemiological studies, and health services research.

Bayes Theorem↗
PubMed2000

Combining contingency tables with missing dimensions.

We propose a methodology for estimating the cell probabilities in a multiway contingency table by combining partial information from a number of studies when not all of the variables are recorded in all studies. We jointly model the full set of categorical variables recorded in at least one of the studies, and we treat the variables that are not reported as missing dimensions of the study-specific contingency table. For example, we might be interested in combining several cohort studies in which the incidence in the exposed and nonexposed groups is not reported for all risk factors in all studies while the overall numbers of cases and cohort size is always available. To account for study-to-study variability, we adopt a Bayesian hierarchical model. At the first stage of the model, the observation stage, data are modeled by a multinomial distribution with fixed total number of observations. At the second stage, we use the logistic normal (LN) distribution to model variability in the study-specific cells' probabilities. Using this model and data augmentation techniques, we reconstruct the contingency table for each study regardless of which dimensions are missing, and we estimate population parameters of interest. Our hierarchical procedure borrows strength from all the studies and accounts for correlations among the cells' probabilities. The main difficulty in combining studies recording different variables is in maintaining a consistent interpretation of parameters across studies. The approach proposed here overcomes this difficulty and at the same time addresses the uncertainty arising from the missing dimensions. We apply our modeling strategy to analyze data on air pollution and mortality from 1987 to 1994 for six U.S. cities by combining six cross-classifications of low, medium, and high levels of mortality counts, particulate matter, ozone, and carbon monoxide with the complication that four of the six cities do not report all the air pollution variables. Our goals are to investigate the association between air pollution and mortality by reconstructing the tables with missing dimensions, to determine the most harmful pollutant combinations, and to make predictions about these key issues for a city other than the six sampled. We find that, for high levels of ozone and carbon monoxide, the number of cases with a high number of deaths increases as the levels of particulate matter, PM10, increases and that the most harmful combinations corresponds to high levels of PM10, confirming prior findings that levels of PM10 higher than the NAAQS standard are harmful.

Air Pollution↗
PubMed1998

A semi-parametric Bayesian approach to generalized linear mixed models.

The linear mixed effects model with normal errors is a popular model for the analysis of repeated measures and longitudinal data. The generalized linear model is useful for data that have non-normal errors but where the errors are uncorrelated. A descendant of these two models generates a model for correlated data with non-normal errors, called the generalized linear mixed model (GLMM). Frequentist attempts to fit these models generally rely on approximate results and inference relies on asymptotic assumptions. Recent advances in computing technology have made Bayesian approaches to this class of models computationally feasible. Markov chain Monte Carlo methods can be used to obtain 'exact' inference for these models, as demonstrated by Zeger and Karim. In the linear or generalized linear mixed model, the random effects are typically taken to have a fully parametric distribution, such as the normal distribution. In this paper, we extend the GLMM by allowing the random effects to have a non-parametric prior distribution. We do this using a Dirichlet process prior for the general distribution of the random effects. The approach easily extends to more general population models. We perform computations for the models using the Gibbs sampler.

Bayes Theorem↗
PubMed1999

Taking account of between-patient variability when modeling decline in Alzheimer's disease.

The pattern of deterioration in patients with Alzheimer's disease is highly variable within a given population. With recent speculation that the apolipoprotein E allele may influence rate of decline and claims that certain drugs may slow the course of the disease, there is a compelling need for sound statistical methodology to address these questions. Current statistical methods for describing decline do not adequately take into account between-patient variability and possible floor and/or ceiling effects in the scale measuring decline, and they fail to allow for uncertainty in disease onset. In this paper, the authors analyze longitudinal Mini-Mental State Examination scores from two groups of Alzheimer's disease subjects from Palo Alto, California, and Minneapolis, Minnesota, in 1981-1993 and 1986-1988, respectively. A Bayesian hierarchical model is introduced as an elegant means of simultaneously overcoming all of the difficulties referred to above.

Alzheimer Disease↗