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 1,711 records · Page 95Linked to original sources
PubMed2004

A log-normal distribution model of the effect of bacteria and ear fenestration on hearing loss: a Bayesian approach.

Chronic ear infection is a potentially life-threatening illness that medical doctors typically treat with ear surgery. Despite the success of this treatment, complications can occur due to bacteria infection. Surgeons believe that this infection causes the patient to have clinically significant hearing damage. In order to understand such complications, surgeons must quantify the effect of bacteria, their toxins and ear surgery on hearing loss. To this end, the other two authors of this paper performed two experiments on guinea pigs to measure hearing thresholds following a bacterial infection and surgery of the inner ear. The response variable in these experiments is hearing thresholds measured in decibels (dB). The problem in analysing such experiments is that the hearing threshold observations often suffer from missing data and censoring mechanisms of various types. Additionally, the distribution of hearing thresholds has heavy tails and is peaked. In order to account for the above statistical issues, we present a Bayesian method with a location-shifted log-normal distribution. The method accounts for the uncertainty in the data collection mechanism and the parameters associated with a location-shifted log-normal distribution. We refer to one of the parameters as the "location-shift" parameter. The Bayesian approach provides a posterior distribution of the location-shift parameter that we compare with values estimated in previously published studies. The immediate goal of our proposed method was to quantify the effects of ear surgery and bacteria infection on hearing loss. Thus, we present the merits of the method in the form of a case study, and report posterior distributions of mean hearing loss, probability of clinically significant hearing loss and relative risk. The results show that surgeon 2, using the surgical procedure "oval window", poses a greater than 40 per cent chance of a 15dB hearing loss regardless of injection of bacteria or not. However, surgeon 1, using the surgical procedure "semicircular canal", does not pose a significantly greater than 40 per cent chance of a 15dB hearing loss unless there is a Pseudomonas aeruginosa-induced infection.

Aged↗
PubMed2000

Physical model-based non-rigid registration incorporating statistical shape information.

This paper describes two new atlas-based methods of 2D single modality non-rigid registration using the combined power of physical and statistical shape models. The transformations are constrained to be consistent with the physical properties of deformable elastic solids in the first method and those of viscous fluids in the second, to maintain smoothness and continuity. A Bayesian formulation, based on each physical model, an intensity similarity measure, and statistical shape information embedded in corresponding boundary points, is employed to derive more accurate and robust approaches to non-rigid registration. A dense set of forces arises from the intensity similarity measure to accommodate complex anatomical details. A sparse set of forces constrains consistency with statistical shape models derived from a training set. A number of experiments were performed on both synthetic and real medical images of the brain and heart to evaluate the approaches. It is shown that statistical boundary shape information significantly augments and improves physical model-based non-rigid registration and the two methods we present each have advantages under different conditions.

Brain↗
PubMed2004

Bayesian estimation of dominance merits in noninbred populations by using Gibbs sampling with two reduced sets of mixed model equations.

Henderson's mixed model equations system is generally required in a Gibbs sampling application. In two previous studies, we proposed two indirect solving approaches that give dominance values in an animal model context with no need to process all this system. The first one does not require D-1 and the second is based on processing the additive animal model residuals. In the present work, we show that these two methods can be handled iteratively. Since the Bayesian approach is now a widely used tool in estimation of genetic parameters, the main part of this work is devoted to a Gibbs sampling application that can be accelerated by means of the aforementioned indirect solving methods. Three replicates of a population data set are simulated in the paper to compare the applications and estimates. This shows effectively that the estimates given by implementing a Gibbs sampler with each of the two suggested solving methods are obtained with less computational time and are comparable to those given by considering the integral system, particularly when priors are more weighted.

Bayes Theorem↗
PubMed1998

Protein construct storage: Bayesian variable selection and prediction with mixtures.

Determining optimal conditions for protein storage while maintaining a high level of protein activity is an important question in pharmaceutical research. A designed experiment based on a space-filling design was conducted to understand the effects of factors affecting protein storage and to establish optimal storage conditions. Different model-selection strategies to identify important factors may lead to very different answers about optimal conditions. Uncertainty about which factors are important, or model uncertainty, can be a critical issue in decision-making. We use Bayesian variable selection methods for linear models to identify important variables in the protein storage data, while accounting for model uncertainty. We also use the Bayesian framework to build predictions based on a large family of models, rather than an individual model, and to evaluate the probability that certain candidate storage conditions are optimal.

Bayes Theorem↗
PubMed2001

Algebraic analysis for nonidentifiable learning machines.

This article clarifies the relation between the learning curve and the algebraic geometrical structure of an unidentifiable learning machine such as a multilayer neural network whose true parameter set is an analytic set with singular points. By using a concept in algebraic analysis, we rigorously prove that the Bayesian stochastic complexity or the free energy is asymptotically equal to lambda(1) log n - (m(1) - 1) log log n + constant, where n is the number of training samples and lambda(1) and m(1) are the rational number and the natural number, which are determined as the birational invariant values of the singularities in the parameter space. Also we show an algorithm to calculate lambda(1) and m(1) based on the resolution of singularities in algebraic geometry. In regular statistical models, 2lambda(1) is equal to the number of parameters and m(1) = 1, whereas in nonregular models, such as multilayer networks, 2lambda(1) is not larger than the number of parameters and m(1) > or = 1. Since the increase of the stochastic complexity is equal to the learning curve or the generalization error, the nonidentifiable learning machines are better models than the regular ones if Bayesian ensemble learning is applied.

Algorithms↗
PubMed2001

Exploring red deer culling strategies using a population-specific calibrated management model.

Wildlife management is generally carried out under conditions of uncertainty. The exact population size is unknown, its future dynamics are uncertain and clear management objectives are often not formulated. In order to provide management advice in this situation, a framework is presented for combining different sources of information using a Bayesian approach for calibrating a management model. Harvesting strategies can then be explored based on predictions of future populations size and structure which incorporate parameter uncertainty. This method makes it possible to evaluate the probability of achieving certain objectives with different management strategies. The advantage of the approach presented in this paper lies in that both the model and the harvesting strategies are adaptable to any particular population of interest. The approach is illustrated for two Scottish red deer populations for which culling strategies corresponding to different management objectives are explored and their benefits evaluated. It is found that each population requires different culling rates for keeping population number stable, demonstrating the benefits of the population specific calibration of the management model.

Animals↗
PubMed2002

Modeling fatigue.

The American Board of Family Practice is developing a patient simulation program to evaluate diagnostic and management skills. The simulator must give temporally and physiologically reasonable answers to symptom questions such as "Have you been tired?" A three-step process generates symptom histories. In the first step, the simulator determines points in time where it should calculate instantaneous symptom status. In the second step, a Bayesian network implementing a roughly physiologic model of the symptom generates a value on a severity scale at each sampling time. Positive, zero, and negative values represent increased, normal, and decreased status, as applicable. The simulator plots these values over time. In the third step, another Bayesian network inspects this plot and reports how the symptom changed over time. This mechanism handles major trends, multiple and concurrent symptom causes, and gradually effective treatments. Other temporal insights, such as observations about short-term symptom relief, require complimentary mechanisms.

Artificial Intelligence↗
PubMed1996

Development and evaluation of a Bayesian pharmacokinetic estimator and optimal, sparse sampling strategies for ceftazidime.

Data were gathered during an activity-controlled trial in which seriously ill, elderly patients were randomized to receive intravenous ceftazidime or ciprofloxacin and for which adaptive feedback control of drug concentrations in plasma and activity profiles was prospectively performed. The adaptive feedback control algorithm for ceftazidime used an initial population model, a maximum a posteriori (MAP)-Bayesian pharmacokinetic parameter value estimator, and an optimal, sparse sampling strategy for ceftazidime that had been derived from data in the literature obtained from volunteers. Iterative two-stage population pharmacokinetic analysis was performed to develop an unbiased MAP-Bayesian estimator and updated optimal, sparse sampling strategies. The final median values of the population parameters were follows: the volume of distribution of the central compartment was equal to 0.249 liter/kg, the volume of distribution of the peripheral compartment was equal to 0.173 liter/kg, the distributional clearance between the central and peripheral compartments was equal to 0.2251 liter/h/kg, the slope of the total clearance (CL) versus the creatinine clearance (CLCR) was equal to 0.000736 liter/h/kg of CL/1 ml/min/1.73 m2 of CLCR, and nonrenal clearance was equal to + 0.00527 liter/h/kg. Optimal sampling times were dependent on CLCR; for CLCR of > or = 30 ml/min/1.73 m2, the optimal sampling times were 0.583, 3.0, 7.0, and 16.0 h and, for CLCR of < 30 ml/min/1.73 m2, optimal sampling times were 0.583, 4.15, 11.5, and 24.0 h. The study demonstrates that because pharmacokinetic information from volunteers may often not be reflective of specialty populations such as critically ill elderly individuals, iterative two-stage population pharmacokinetic analysis, MAP-Bayesian parameter estimation, and optimal, sparse sampling strategy can be important tools in characterizing their pharmacokinetics.

Adult↗
PubMed2005

Random effects survival models gave a better understanding of heterogeneity in individual patient data meta-analyses.

BACKGROUND AND OBJECTIVE: Individual patient data meta-analysis consists in combining data from all available trials dealing with a therapeutic problem in order to increase the power of statistical analyses. A key issue when analyzing these pooled data sets is intertrial heterogeneity. In survival data, heterogeneity manifests itself either by differing treatment effects between the included trials or by a baseline hazard that differs between studies. One way to investigate and accommodate this heterogeneity is to use models that include random effects. METHODS: We apply this class of models to the Meta-Analysis of Chemotherapy in Head and Neck Cancers, in which strong heterogeneity is exhibited. This meta-analysis pooled 63 trials involving 10,741 patients. RESULTS: We show that such modeling permits a better understanding of heterogeneity in the MACH-NC data, both from a frequentist and from a Bayesian point of view. In particular, the modeling suggests the presence of two outlying sets of trials whose baseline risk could explain the apparent efficacy or inefficacy of some treatment protocols. CONCLUSION: We conclude that this family of random-effects models is a useful tool for exploring heterogeneity in meta-analyses of time-to-event data, and that its features can be applied to a very wide range of studies.

Chemotherapy, Adjuvant↗
PubMed2005

Easy-to-use, accurate and flexible individualized Bayesian limited sampling method without fixed time points for ciclosporin monitoring after liver transplantation.

BACKGROUND: New methods to estimate the systemic exposure to ciclosporin such as the level 2 h after dosing and limited sampling formulas may lead to improved clinical outcome after orthotopic liver transplantation. However, most strategies are characterized by rigid sampling times. AIM: To develop and validate a flexible individualized population-pharmacokinetic model for ciclosporin monitoring in orthotopic liver transplantation. METHODS: A total of 62 curves obtained from 31 patients at least 0.5 year after orthotopic liver transplantation were divided into two equal groups. From 31 curves, relatively simple limited sampling formulas were derived using multiple regression analysis, while using pharmacokinetic software a two-compartment population-pharmacokinetic model was derived from these same data. We then tested the ability to estimate the AUC by the limited sampling formulas and a different approach using several limited sampling strategies on the other 31 curves. The new approach consists of individualizing the mean a priori population-pharmacokinetic parameters of the two-compartment population-pharmacokinetic model by means of maximum a posteriori Bayesian fitting with individual data leading to an individualized population-pharmacokinetic limited sampling model. From the individualized pharmacokinetic parameters, AUC(0-12h) was calculated for each combination of measured blood concentrations. The calculated AUC(0-12h) both from the limited-sampling formulas and the limited-sampling model were compared with the gold standard AUC(0-12h) (trapezoidal rule) by Pearson's correlation coefficient and prediction precision and bias were calculated. RESULTS: The AUC(0-12h) value calculated by individualizing the population-pharmacokinetic model using several combinations of measured blood concentrations: 0 + 2 h (r(2) = 0.94), 0 + 1 + 2 h (r(2) = 0.94), 0 + 1 + 3 h (r(2) = 0.92), 0 + 2 + 3 h (r(2) = 0.92) and 0 + 1 + 2 + 3 h (r(2) = 0.96) had excellent correlation with AUC(0-12h), better than limited sampling formulas with less than three sampling time points. Even trough level with limited sampling method (r(2) = 0.86) correlated better than the level after 2 h of dosing (r(2) = 0.75) or trough level (r(2) = 0.64) as single values without limited sampling method. Moreover, the individualized population-pharmacokinetic model had a low prediction bias and excellent precision. CONCLUSION: Multiple rigid sampling time points limit the use of limited sampling formulas. The major advantage of the Bayesian estimation approach presented here, is that blood sampling time points are not fixed, as long as sampling time is known. The predictive performance of this new approach is superior to trough level and that after 2 h of dosing and at least as good as limited sampling formulas. It is of clear advantage in busy out-patient clinics.

Adult↗
PubMed2001

Bayesian analysis of liability of clinical mastitis in Norwegian cattle with a threshold model: effects of data sampling method and model specification.

First-lactation records of Norwegian Cattle were used to infer heritability of liability to clinical mastitis with a threshold sire model. Mastitis was defined as a binary response (presence or absence) in a defined period of first lactation (opportunity period). Length of opportunity period (from 30 d before calving up to 120 or 300 d of lactation) had less effect on heritability estimates than data sampling methods (include or exclude records of cows culled before the end of the opportunity period) whereas sire ranking was more affected by the former. Including all cows, whether culled before the end of the opportunity period or not, gave a sharper and more symmetric posterior distribution of heritability of liability to clinical mastitis. When we analyzed data for all cows, model specification had a small effect on heritability estimates, while sire ranking was affected markedly. Posterior means of heritability range from 0.058 to 0.074. A model regressing on the length of the opportunity period for culled cows without mastitis, was shown favorable for the two opportunity periods using Bayes factors and the deviance information criterion for model comparison. This model, in which liability of mastitis depends on time to culling, may allow utilizing information from all first lactations in genetic evaluation, irrespectively of duration and culling outcome.

Animals↗
PubMed2002

Probabilistic Boolean Networks: a rule-based uncertainty model for gene regulatory networks.

MOTIVATION: Our goal is to construct a model for genetic regulatory networks such that the model class: (i) incorporates rule-based dependencies between genes; (ii) allows the systematic study of global network dynamics; (iii) is able to cope with uncertainty, both in the data and the model selection; and (iv) permits the quantification of the relative influence and sensitivity of genes in their interactions with other genes. RESULTS: We introduce Probabilistic Boolean Networks (PBN) that share the appealing rule-based properties of Boolean networks, but are robust in the face of uncertainty. We show how the dynamics of these networks can be studied in the probabilistic context of Markov chains, with standard Boolean networks being special cases. Then, we discuss the relationship between PBNs and Bayesian networks--a family of graphical models that explicitly represent probabilistic relationships between variables. We show how probabilistic dependencies between a gene and its parent genes, constituting the basic building blocks of Bayesian networks, can be obtained from PBNs. Finally, we present methods for quantifying the influence of genes on other genes, within the context of PBNs. Examples illustrating the above concepts are presented throughout the paper.

Cell Cycle↗
PubMed1999

The Rose model, revisited.

In 1946 and 1948, three very important papers by Albert Rose [J. Soc. Motion Pict. Eng. 47, 273 (1946); J. Opt. Soc. Am. 38, 196 (1948); L. Marton, ed. (Academic, New York, 1948)] were published on the role that photon fluctuations have in setting fundamental performance limits for both human vision and electronic imaging systems. The papers were important because Rose demonstrated that the performance of imaging devices can be evaluated with an absolute scale (quantum efficiency). The analysis of human visual signal detection used in these papers (developed before the formal theory of signal detectability) was based on an approach that has come to be known as the Rose model. In spite of its simplicity, the Rose model is a very good approximation of a Bayesian ideal observer for the carefully and narrowly defined conditions that Rose considered. This simple model can be used effectively for back-of-the-envelope calculations, but it needs to be used with care because of its limited range of validity. One important conclusion arising from Rose's investigations is that pixel signal-to-noise ratio is not a good figure of merit for imaging systems or components, even though it is still occasionally used as such by some researchers. In the present study, (1) aspects of signal detection theory are presented, (2) Rose's model is described and discussed, (3) pixel signal-to-noise ratio is discussed, and (4) progress on modeling human noise-limited performance is summarized. This study is intended to be a tutorial with presentation of the main ideas and provision of references to the (dispersed) technical literature.

Artifacts↗
PubMedDate not supplied

Prospective validation of an optimal sparse plasma-sampling strategy for estimating ciprofloxacin pharmacokinetics.

Data obtained from 23 critically ill patients treated with intravenous ciprofloxacin in two clinical trials were used to validate prospectively a previously developed maximum a posteriori (MAP)-Bayesian estimator and optimal plasma-sampling strategy (OSS). Dosages ranged from 200 mg every 12 hours to 400 mg every 8 hours. Each patient had 8-16 samples taken, either as large gold standard sampling sets or as a mix of gold standard sets and OSSs. The MAP-Bayesian estimator used a two-compartment model and identified apparent volumes of distribution of the central and peripheral compartments, distributional clearance, and the slope and intercept of the relationship between creatinine clearance and total body clearance. Fit parameters were used to derive the apparent volume of distribution at steady state and the 24-hour area under the curve. All parameters derived from the OSS using the MAP-Bayesian estimates matched up almost identically to those obtained from modeling the gold standard sets. There was no systematic bias, and good precision was seen among all the parameters. These data demonstrate the usefulness and validity of the current OSS and MAP-Bayesian estimator, and provide further evidence of the utility of optimal sampling theory.

Adult↗
PubMed1997

Advances in statistical methods to map quantitative trait loci in outbred populations.

Statistical methods to map quantitative trait loci (QTL) in outbred populations are reviewed, extensions and applications to human and plant genetic data are indicated, and areas for further research are identified. Simple and computationally inexpensive methods include (multiple) linear regression of phenotype on marker genotypes and regression of squared phenotypic differences among relative pairs on estimated proportions of identity-by-descent at a locus. These methods are less suited for genetic parameter estimation in outbred populations but allow the determination of test statistic distributions via simulation or data permutation; however, further inferences including confidence intervals of QTL location require the use of Monte Carlo or bootstrap sampling techniques. A method which is intermediate in computational requirements is residual maximum likelihood (REML) with a covariance matrix of random QTL effects conditional on information from multiple linked markers. Testing for the number of QTLs on a chromosome is difficult in a classical framework. The computationally most demanding methods are maximum likelihood and Bayesian analysis, which take account of the distribution of multilocus marker-QTL genotypes on a pedigree and permit investigators to fit different models of variation at the QTL. The Bayesian analysis includes the number of QTLs on a chromosome as an unknown.

Bayes Theorem↗
PubMed1991

The prediction of steady-state plasma phenobarbitone concentrations (following low-dose phenobarbitone) to refine its use as an indicator of compliance.

1. A model for predicting the steady-state plasma concentration of phenobarbitone following low-dose phenobarbitone used as an indicator of compliance was derived using data for 10 healthy volunteers. 2. Each volunteer was given a single 30 mg oral dose of phenobarbitone and the pharmacokinetics were described. Subsequently, volunteers were given phenobarbitone 2 mg daily for 28 days and a further pharmacokinetic profile determined during and after this period. 3. An initial predicted estimate of steady-state plasma drug concentration was made using each volunteer's demographic details. This estimate was revised by Bayesian analysis using single timed samples (24, 48, 72 or 96 h) following the single dose. 4. The model was tested on a further 10 healthy volunteers given a single 8 mg dose and who were subsequently given 2 mg daily for 28 days. 5. The revised estimate of peak steady-state plasma phenobarbitone concentration utilising the 96 h post-single dose concentration (356 ng ml-1) was least biased (mean prediction error +/- 95% CI = 10.6 +/- 19.8 ng ml-1) and most precise (root mean square error +/- 95% CI = 28.3 +/- 19.0 ng ml-1). In all cases the peak or trough steady-state drug concentration was within 13% of the predicted value. 6. The model reflected compliance accurately in a further eight volunteers with simulated partial (two-thirds) compliance. 7. The use of a predictive model using Bayesian analysis to estimate expected steady-state plasma phenobarbitone concentrations could increase further the usefulness of low-dose phenobarbitone as an indicator of compliance.

Adult↗
PubMed2004

Hunting drug targets by systems-level modeling of gene expression profiles.

Structural learning of Bayesian networks applied to sets of genome-wide expression patterns has been recently discovered as a potentially useful tool for the systems-level statistical description of gene interactions. We train and analyze Bayesian networks with the goal of inferring biological aspects of gene function. Our two-component approach focuses on supporting the drug discovery process by identifying genes with central roles for the network operation, which could act as drug targets. The first component, referred to as scale-free analysis, uses topological measures of the network-related to a high-traffic load of genes-as estimators for their functional importance. The second component, referred to as generative inverse modeling, is a method of estimating the effect of a simulated drug treatment or mutation on the global state of the network, as measured in the expression profile. We show for a dataset from acute lymphoblastic leukemia patients that both approaches are suitable for finding genes with central cellular functions. In addition, generative inverse modeling correctly identifies a known oncogene in a purely data-driven way.

Algorithms↗