SEARCH · PubMed Health

Results for “Bayesian analysis”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 487 records · Page 27Linked to original sources

PubMed2003

Language-tree divergence times support the Anatolian theory of Indo-European origin.

Languages, like genes, provide vital clues about human history. The origin of the Indo-European language family is "the most intensively studied, yet still most recalcitrant, problem of historical linguistics". Numerous genetic studies of Indo-European origins have also produced inconclusive results. Here we analyse linguistic data using computational methods derived from evolutionary biology. We test two theories of Indo-European origin: the 'Kurgan expansion' and the 'Anatolian farming' hypotheses. The Kurgan theory centres on possible archaeological evidence for an expansion into Europe and the Near East by Kurgan horsemen beginning in the sixth millennium BP. In contrast, the Anatolian theory claims that Indo-European languages expanded with the spread of agriculture from Anatolia around 8,000-9,500 years bp. In striking agreement with the Anatolian hypothesis, our analysis of a matrix of 87 languages with 2,449 lexical items produced an estimated age range for the initial Indo-European divergence of between 7,800 and 9,800 years bp. These results were robust to changes in coding procedures, calibration points, rooting of the trees and priors in the bayesian analysis.

Agriculture↗

PubMed1996

Invasive cervical carcinoma: role of MR imaging in pretreatment work-up--cost minimization and diagnostic efficacy analysis.

PURPOSE: To examine the cost and efficacy of diagnostic work-up in patients with invasive cervical cancer. MATERIALS AND METHODS: In 246 patients with invasive cervical cancer, all diagnostic tests performed before treatment were recorded. Patients were divided into two groups: those who underwent magnetic resonance (MR) imaging as the initial study (n = 105) and those who did not (n = 141). A list of 1995 Medicare global payments was used to measure cost. Bayesian analysis (likelihood ratios derived from a literature search) was performed for bladder, rectal, parameterial, and nodal involvement in stage Ib disease. RESULTS: Significantly fewer procedures and fewer invasive studies were performed in the MR imaging group. Net cost savings for the MR imaging group was $401 for all patients and $449 for patients with stage Ib disease. For stage Ib disease, the 0% pretest probability of bladder or rectal invasion does not justify the routine use of barium enema examination, cystoscopy, or proctoscopy. The increase in predictive values for parameterial and nodal disease was highest for MR imaging when tumor size was at least 2 cm. CONCLUSION: Guidelines for the pretreatment work-up of clinical stage Ib cervical cancer need revision. MR imaging should be used as an adjunct to clinical evaluation.

Adenocarcinoma↗

PubMed2003

Assessing and comparing costs: how robust are the bootstrap and methods based on asymptotic normality?

This article addresses and challenges some common perceptions in the statistical assessment of costs and cost-effectiveness in health economics. Cost data typically exhibit highly skew distributions. Two techniques whose validity does not depend on any specific form of underlying distribution are the bootstrap and methods based on asymptotic normality of sample means. These methods are generally thought to be appropriate for the analysis of cost data. We argue that, even when these methods are technically valid, they may often lead to inefficient and even misleading inferences. It is important to apply methods that recognise the skewness in cost data. We further demonstrate that it may also be important to incorporate relevant prior information in a Bayesian analysis.

Bayes Theorem↗

PubMed2004

Model parameterization, prior distributions, and the general time-reversible model in Bayesian phylogenetics.

Bayesian phylogenetic methods require the selection of prior probability distributions for all parameters of the model of evolution. These distributions allow one to incorporate prior information into a Bayesian analysis, but even in the absence of meaningful prior information, a prior distribution must be chosen. In such situations, researchers typically seek to choose a prior that will have little effect on the posterior estimates produced by an analysis, allowing the data to dominate. Sometimes a prior that is uniform (assigning equal prior probability density to all points within some range) is chosen for this purpose. In reality, the appropriate prior depends on the parameterization chosen for the model of evolution, a choice that is largely arbitrary. There is an extensive Bayesian literature on appropriate prior choice, and it has long been appreciated that there are parameterizations for which uniform priors can have a strong influence on posterior estimates. We here discuss the relationship between model parameterization and prior specification, using the general time-reversible model of nucleotide evolution as an example. We present Bayesian analyses of 10 simulated data sets obtained using a variety of prior distributions and parameterizations of the general time-reversible model. Uniform priors can produce biased parameter estimates under realistic conditions, and a variety of alternative priors avoid this bias.

Animals↗

PubMed1991

Three-locus linkage analysis using recombinant inbred strains and Bayes' theorem.

Recombinant inbred (RI) strains are useful in linkage analysis and gene mapping. However, the generally small number of strains in an RI strain set limits the power of RI strains in linkage detection. Several methods for increasing the power of RI strains have been used, including summing data across RI strain sets and excluding linkage to genomic regions. In this paper, Bayesian analysis is applied to three-locus linkage data. This method further increases the power of RI strains to detect linkage and gives estimations of the probability of each of the three possible gene orders if the test locus is linked to the pair of marker loci. Several examples are presented, including reconsideration of the position of the proto-oncogene L-myc on the mouse linkage map.

Animals↗

PubMed2005

Molecular analysis reveals tighter social regulation of immigration in patrilocal populations than in matrilocal populations.

Human social organization can deeply affect levels of genetic diversity. This fact implies that genetic information can be used to study social structures, which is the basis of ethnogenetics. Recently, methods have been developed to extract this information from genetic data gathered from subdivided populations that have gone through recent spatial expansions, which is typical of most human populations. Here, we perform a Bayesian analysis of mitochondrial and Y chromosome diversity in three matrilocal and three patrilocal groups from northern Thailand to infer the number of males and females arriving in these populations each generation and to estimate the age of their range expansion. We find that the number of male immigrants is 8 times smaller in patrilocal populations than in matrilocal populations, whereas women move 2.5 times more in patrilocal populations than in matrilocal populations. In addition to providing genetic quantification of sex-specific dispersal rates in human populations, we show that although men and women are exchanged at a similar rate between matrilocal populations, there are far fewer men than women moving into patrilocal populations. This finding is compatible with the hypothesis that men are strictly controlling male immigration and promoting female immigration in patrilocal populations and that immigration is much less regulated in matrilocal populations.

Bayes Theorem↗

PubMedDate not supplied

Constructing Bayesian formulations of sparse kernel learning methods.

We present here a simple technique that simplifies the construction of Bayesian treatments of a variety of sparse kernel learning algorithms. An incomplete Cholesky factorisation is employed to modify the dual parameter space, such that the Gaussian prior over the dual model parameters is whitened. The regularisation term then corresponds to the usual weight-decay regulariser, allowing the Bayesian analysis to proceed via the evidence framework of MacKay. There is in addition a useful by-product associated with the incomplete Cholesky factorisation algorithm, it also identifies a subset of the training data forming an approximate basis for the entire dataset in the kernel-induced feature space, resulting in a sparse model. Bayesian treatments of the kernel ridge regression (KRR) algorithm, with both constant and heteroscedastic (input dependent) variance structures, and kernel logistic regression (KLR) are provided as illustrative examples of the proposed method, which we hope will be more widely applicable.

Algorithms↗

PubMed2006

Prognosis of multiple sclerosis: clinical factors predicting the late evolution for an early treatment decision.

With the development of new immunomodulating therapies, with which early use is strongly encouraged, it is crucial to be in possession of reliable clinical predictors of multiple sclerosis evolution. Prognostic factors are important to patients wanting to be informed about their prospects; to the clinician needing to individualize patients requiring immune treatments at an early stage of the disease; and also to the researcher needing to to improve the design and analysis of the clinical therapeutic trials and observational studies. Frequentist analyses have indicated a poor prognosis for male gender, late age at onset, motor, cerebellar and sphincter involvement at onset, progressive course at onset, short inter-attack interval, high number of early attacks; and a relevant early residual disability. A recent application of a Bayesian analysis led to the construction of more detailed models of the natural history of multiple sclerosis and the estimated risk of unfavorable evolution at an individual patient level.

Age Factors↗

PubMed1990

Length of psychiatric hospitalization and prediction of antipsychotic response.

1. Clinical variables determining length of psychiatric hospitalization for psychotic inpatients were explored. Forty psychotic inpatients received a 14 day fixed dose neuroleptic trial. 2. Neuroleptic responders (25/40) were discharged 15 +/- 2 days after initiation of pharmacotherapy. For neuroleptic non-responders (15/40) antipsychotic medication was then altered as clinically indicated. Patients requiring one change in medication (N = 8) were discharged after 27 +/- 5 days; those requiring two medication adjustments (N = 4) were discharged after 33 +/- 3 days and those requiring three alterations in pharmacotherapy (N = 3) were discharged after 42 +/- 12 days. 3. Statistical analysis of clinical and diagnostic variables indicated that 84% of the variation in length of hospitalization was accounted for by the number of alterations in pharmacotherapy required for symptom remission and discharge. It is suggested that length of hospitalization may be decreased by decreasing the length of time a clinician prescribes pharmacotherapy that subsequently proves not effective. 4. Bayesian analysis was employed to identify the minimum length of pharmacotherapy which accurately predicts subsequent antipsychotic response/non-response. During the fixed dose neuroleptic trial response/non-response could be accurately predicted for 65% of the patients by Day 3 of the trial while by Day 7 response/non-response could be predicted for 80% of the patients. 5. The present data indicate that a three to seven day trial of antipsychotics may be sufficient for making pharmacotherapy decisions as such a trial demonstrates a diagnostic efficiency similar to other predictive tests employed in clinical medicine.

Fluphenazine↗

PubMed2006

Functional analysis of gene duplications in Saccharomyces cerevisiae.

Gene duplication can occur on two scales: whole-genome duplications (WGD) and smaller-scale duplications (SSD) involving individual genes or genomic segments. Duplication may result in functionally redundant genes or diverge in function through neofunctionalization or subfunctionalization. The effect of duplication scale on functional evolution has not yet been explored, probably due to the lack of global knowledge of protein function and different times of duplication events. To address this question, we used integrated Bayesian analysis of diverse functional genomic data to accurately evaluate the extent of functional similarity and divergence between paralogs on a global scale. We found that paralogs resulting from the whole-genome duplication are more likely to share interaction partners and biological functions than smaller-scale duplicates, independent of sequence similarity. In addition, WGD paralogs show lower frequency of essential genes and higher synthetic lethality rate, but instead diverge more in expression pattern and upstream regulatory region. Thus, our analysis demonstrates that WGD paralogs generally have similar compensatory functions but diverging expression patterns, suggesting a potential of distinct evolutionary scenarios for paralogs that arose through different duplication mechanisms. Furthermore, by identifying these functional disparities between the two types of duplicates, we reconcile previous disputes on the relationship between sequence divergence and expression divergence or essentiality.

Bayes Theorem↗

PubMed1994

Hierarchical regression analysis applied to a study of multiple dietary exposures and breast cancer.

Hierarchical regression attempts to improve standard regression estimates by adding a second-stage "prior" regression to an ordinary model. Here, we use hierarchical regression to analyze case-control data on diet and breast cancer. This regression yields semi-Bayes relative risk estimates for dietary items by using a second-stage model to pull estimates toward each other when the corresponding variables have similar levels of nutrients. Unlike classical Bayesian analysis, however, no use is made of previous studies on nutrient effects. Compared with results obtained with one-stage conditional maximum-likelihood logistic regression, our hierarchical regression model gives more stable and plausible estimates. In particular, certain effects with implausible maximum-likelihood estimates have more reasonable semi-Bayes estimates.

Bayes Theorem↗

PubMed2004

Prevalence of spondylosis deformans and estimates of genetic parameters for the degree of osteophytes development in Italian Boxer dogs.

The aim of this study was to assess the prevalence of spondylosis deformans and to investigate genetic aspects of the degree of osteophytes development (DOD) in the Italian Boxer dog population. A total of 849 Boxer dogs was radiographed on the thoracic, lumbar, and sacral regions of the spine and scored for DOD. Grading of DOD was performed for all 20 intervertebral sites comprised within the first thoracic site (site T1-T2) and the site between the seventh lumbar and the first sacral vertebra (site L7-S1). Scores for DOD ranged from 0 (no osteophytes development) to 3 (presence of a bony spur formed by osteophytes on adjoining vertebrae). The first five thoracic sites exhibited no variation for DOD and were not considered in the analysis. The prevalence of spondylosis deformans was 84%, and frequency of dogs showing at least one intervertebral site that scored 3 for DOD was 50%. Scores for DOD at different sites were analyzed as different traits. Nongenetic effects influencing DOD scores were sex, age at screening, and the kennel. Posterior densities of heritability (h2) were estimated using a univariate Bayesian analysis. Eight sites exhibited a posterior probability greater than 0.8 for h2 > 10% and were considered in a multivariate restricted maximum likelihood analysis. Estimated h2 from multivariate analysis ranged from 25 to 48% (SE from 5 to 7%). Three sites exhibited h2 estimates greater than 40%. Genetic correlations for DOD scored at different sites ranged from 0.07 to 0.96. All thoracic sites had estimated correlations larger than 0.85 with other thoracic sites. Genetic correlation between the first and the second lumbar site was 0.91. Correlations between thoracic sites and the first two lumbar sites ranged from 0.5 to 0.9. Sites L6-L7 and L7-S1 also exhibited weak relationships with all remaining sites. Breeding values of dogs for DOD at the eight sites were predicted using estimated covariance matrices. A selection index for DOD was computed from predicted breeding values and a set of relative weighting factors produced by a panel of veterinarians. The index was the most important effect influencing phenotypic differences between dogs for average DOD score, number of affected sites, and number of sites with a DOD score > 1 (P < 0.001). The degree of osteophytes development is a trait showing exploitable additive genetic variance, and breeding programs for decreasing prevalence and severity of spondylosis deformans might focus on this trait.

Age Factors↗

PubMed1978

General problems in medical decision making with comments on ROC analysis.

Medical decision-making studies continue to focus on two questions: How do physicians make decisions? How should physicians make decisions? Researchers pursuing the first question emphasize human cognitive processes and the programming of symbol systems to model observed human behavior. Those researchers concentrating on the second question assume that there is a standard of performance against which the physician's decisions can be judged, and to help the physician improve his performance, an array of tools is proposed. These tools include decision trees, Bayesian analysis, decision matrices, receiver operating characteristics (ROC) analysis, and cost-benefit considerations including utility measures. Medical decision-making questions must be answered in an ethical context where ethics and decision analysis are interviewed.

Bayes Theorem↗

PubMed2004

Rapid radiation and cryptic speciation in squat lobsters of the genus Munida (Crustacea, Decapoda) and related genera in the South West Pacific: molecular and morphological evidence.

Squat lobsters (genus Munida and related genera) are among the most diverse taxa of western Pacific crustaceans, though several features of their biology and phylogenetic relationships are unknown. This paper reports an extensive phylogenetic analysis based on mitochondrial DNA sequences (cytochrome c oxidase subunit I and 16S rRNA) and the morphology of 72 species of 12 genera of western Pacific squat lobsters. Our phylogenetic reconstruction using molecular data supports the recent taxonomic splitting of the genus Munida into several genera. Excluding one species (M. callista), the monophyly of the genus Munida was supported by Bayesian analysis of the molecular data. Three moderately diverse genera (Onconida, Paramunida, and Raymunida) also appeared monophyletic, both according to morphological and molecular data, always with high support. However, other genera (Crosnierita and Agononida) seem to be para- or polyphyletic. Three new cryptic species were identified in the course of this study. It would appear that the evolution of this group was marked by rapid speciation and stasis, or certain constraints, in its morphological evolution.

Animals↗

PubMed2005

The utility of prior information and stratification for parameter estimation with two screening tests but no gold standard.

When a gold standard screening or diagnostic test is not routinely available, it is common to apply two different imperfect tests to subjects from a study population. There is a considerable literature on estimating relevant parameters from the resultant data. In the situation that test sensitivities and specificities are unknown, several inferential strategies have been proposed. One suggestion is to use rough knowledge about the unknown test characteristics as prior information in a Bayesian analysis. Another suggestion is to obtain the statistical advantage of an identified model by splitting the population into two strata with differing disease prevalences. There is some division of opinion in the epidemiological literature on the relative merits of these two approaches. This article aims to shed light on the issue, by applying some recently developed theory on the performance of Bayesian inference in non-identified statistical models.

Bayes Theorem↗

PubMed2006

Evaluation of three serological tests for diagnosis of Maedi-Visna virus infection using latent class analysis.

Maedi-Visna virus (MVV) infection in sheep is present in several European countries, including Norway. The current Norwegian surveillance and control programme for MVV infection uses three serological tests: an agar gel immunodiffusion test (AGID) and two commercially available indirect ELISAs (Institut Pourquier, P-ELISA and HYPHEN BioMed, H-ELISA). From 18 flocks with suspected or confirmed MVV infection, sera from naturally infected sheep were obtained, and sensitivity (Se) and specificity (Sp) of the three tests were estimated in absence of a perfect reference test using latent class models in a Bayesian analysis. The AGID had higher Sp (95% posterior credibility interval (PCI) [98.4; 99.9]) than either ELISA (95% PCI: P-ELISA, [95.1; 99.0]; H-ELISA, [91.4; 96.6]), but much lower Se (95% PCI: AGID, [41.4; 59.8]; P-ELISA, [92.7; 100.0]; H-ELISA, [90.9; 99.4]). Currently the P-ELISA is used for screening and positive samples are subsequently confirmed by a setup using all three tests in a serial reading. The Se and Sp of the serial interpretations with and without the H-ELISA were estimated. The results suggested that the H-ELISA could be dropped as a confirmatory test as the Se of the three test serial reading was reduced significantly without adding a significant improvement of the Sp compared to the serial reading of the P-ELISA and AGID alone. However, the perceived cost of false positives versus false negatives will influence this decision. Estimates of the predictive values for the tests and combinations suggested that the P-ELISA is a good choice of screening, but confirmatory tests are needed to achieve acceptable levels of positive predictive values.

Animals↗

PubMed1998

Modelling behavioral syndromes using Bayesian networks.

In this paper Bayesian networks modelling is applied to a multidimensional model of depression. The characterization of the probabilistic model exploits expert knowledge to associate latent concentrations of neurotransmitters and symptoms. An evolution perspective is also considered. Specific criteria are introduced to detect the influence of the latent variable on the observation of symptoms. The Bayesian analysis is carried out using Gibbs sampling technique which is implemented in the BUGS software. The estimation phase leads to the selection of symptoms entering into the definition of behavioral syndromes. Results on real data are discussed. The last section deals with simulation experiments. Simulation results confirm our methodological choices. Results of the paper can enlarge to the central problem of the management of latent variables in Bayesian networks modelling.

Artificial Intelligence↗

PubMed2006

Prior specification in Bayesian statistics: three cautionary tales.

One of the most important differences between Bayesian and traditional techniques is that the former combines information available beforehand-captured in the prior distribution and reflecting the subjective state of belief before an experiment is carried out-and what the data teach us, as expressed in the likelihood function. Bayesian inference is based on the combination of prior and current information which is reflected in the posterior distribution. The fast growing implementation of Bayesian analysis techniques can be attributed to the development of fast computers and the availability of easy to use software. It has long been established that the specification of prior distributions should receive a lot of attention. Unfortunately, flat distributions are often (inappropriately) used in an automatic fashion in a wide range of types of models. We reiterate that the specification of the prior distribution should be done with great care and support this through three examples. Even in the absence of strong prior information, prior specification should be done at the appropriate scale of biological interest. This often requires incorporation of (weak) prior information based on common biological sense. Very weak and uninformative priors at one scale of the model may result in relatively strong priors at other levels affecting the posterior distribution. We present three different examples intuïvely illustrating this phenomenon indicating that this bias can be substantial (especially in small samples) and is widely present. We argue that complete ignorance or absence of prior information may not exist. Because the central theme of the Bayesian paradigm is to combine prior information with current data, authors should be encouraged to publish their raw data such that every scientist is able to perform an analysis incorporating his/her own (subjective) prior distributions.

Animals↗