Search

In this paper, we present a method to assess differences between microbiome communities that effectively models sparse count data and accounts for presence-absence bias frequently encountered when zeros are present. We assume that the observed data for each operational taxonomic unit is Poisson generated with the rate for each sample originating from an underlying rate distribution. We propose to model this distribution using a mixture model, specifying the components based on the posterior rate distribution of a count and estimating the optimal weights using a least squares objective function. The distribution incorporates varying resolutions of samples, a point mass for differentiating structural and nonstructural zeros, and a truncation point mass to account for high values that are too sparse to model. As mixture component specification is not always straightforward, a method to estimate a joint model from several mixture distributions using minimum distances of bootstrap iterates is proposed. Once the population rate distribution is approximated, we obtain sample-specific distributions by conditioning on the observed operational taxonomic unit count, resolution, and estimated mixture distribution and then use these to estimate pairwise distances for a permutation test. The method gives an accurate estimate of the true proportion of zeros for presence-absence, effectively models the distribution of low counts using the mixture distribution, and achieves good power for detecting differences in a variety of scenarios. The method is tested using a simulation study and applied to two microbiome datasets. In each case, the results are compared with a number of existing methods. [ABSTRACT FROM AUTHOR]

Physiological performance in fish peaks within a well-defined range of temperatures, which is distinct for each species. Species-specific thermal responses for growth, survival, and reproduction are most commonly quantified directly through laboratory experiment or field observation, with a focus on six specific metrics: optimum growth temperature and final temperature preferendum (growth), upper incipient lethal temperature and critical thermal maximum (survival), and optimum spawning temperature and optimum egg development temperature (reproduction). These values remain unknown for many North American freshwater fish species. In this paper, we present a new statistical method (Bayesian phylogenetic regression) that uses relationships between these metrics and phenetic relatedness to estimate unknown metric values. The reliability of these estimates was compared with those derived from models incorporating taxonomic family and models without any taxonomic information. Overall, incorporating taxonomic family relatedness improved estimation accuracy across all metrics. For Salmonidae and Cyprinidae, estimates derived from Bayesian phylogenetic regression typically had the highest expected reliability. We used our methods to generate 274 estimates of unknown metric values for over 100 North American freshwater fish species. [ABSTRACT FROM AUTHOR]

Rock engineering design is currently evolving from the traditional use of safety factors and towards the more rational reliability-based design (RBD), as witnessed by the introduction of a number of geotechnical limit states design (LSD) standards worldwide. The probabilistic nature of RBD requires statistical characterization of design parameters, including the strength of intact rock. The triaxial compressive strength of intact rock is commonly characterized as a function of confining pressure, and values of the parameters that define this function are generally obtained by regression analysis of laboratory test data. However, the problem of fitting strength criteria to intact rock strength data has been historically tackled as a problem of obtaining best fit curves only, and has inadvertently omitted characterization of the variability inherent in the strength data. As a result, the uncertainty in parameter estimations resulting from this variability are often not quantified rigorously. Not only does this omission render such regression analyses incomplete, it also limits development of reliability-based design protocols for rock engineering which require statistical characterization of design parameters. To overcome both of these deficiencies, this paper presents frequentist (i.e. classical) and Bayesian regression models that rigorously incorporate variability and uncertainty associated with estimated Hoek-Brown strength parameters. In particular, it discusses the limitations of the frequentist model when dealing with limited data or a combination of tensile and compressive strength data. It discusses the potential of Bayesian data analysis techniques to overcome the issue of limited data in rock engineering design by using informative prior distributions. The paper also demonstrates the main issue in fitting strength criteria to tensile and compressive data simultaneously from a statistical perspective, and presents a Bayesian regression model that rigorously fits the strength criterion to a combination of tensile and compressive data. The paper concludes with suggested statistical approaches – Bayesian or frequentist – for conditions encountered in the analysis of intact rock strength data. [ABSTRACT FROM AUTHOR]

Finite mixture models are flexible parametric models that allow one to describe complex probability distributions as a mixture of a small number of simple probability distributions. Measures of health status are often used to reflect a person's overall health. Such measures may be subject to a ceiling effect, in that the measure is unable to discern gradations in health status above the ceiling. The purpose of this paper is to illustrate the use of finite mixture models to describe the probability distribution of the Health Utilities Index, under the assumption that the HUI is subject to a ceiling effect. Mixture models with two through six components are fit to the HUI. Bayes factors were used to compare the evidence that the Canadian population of non-institutionalized residents is composed of four distinct subpopulations, and that a mixture of six Normal components is required to describe these four subpopulations. [ABSTRACT FROM AUTHOR]

We describe and illustrate Bayesian inference in models for density estimation using mixtures of Dirichlet processes. These models provide natural settings for density estimation and are exemplified by special cases where data are modeled as a sample from mixtures of normal distributions. Efficient simulation methods are used to approximate various prior, posterior, and predictive distributions. This allows for direct inference on a variety of practical issues, including problems of local versus global smoothing, uncertainty about density estimates, assessment of modality, and the inference on the numbers of components. Also, convergence results are established for a general class of normal mixture models. [ABSTRACT FROM AUTHOR]

Uses the Dirichlet process prior to provide a nonparametric Bayesian estimate of a vector of normal means. Development of a `Gibbs sampler' algorithm; Two classes of empirical Bayes method; Discussion of ways to estimate the vector Xi's.

With the introduction and revisions of geotechnical limit states design (LSD) standards such as Eurocode 7, rock engineering design is moving towards reliability-based design, a method for which statistical characterisation of design parameters is essential. However, the often limited project-specific data in rock engineering do not allow straightforward application of classical statistical analyses, and thus alternative approaches are required. In this paper, hierarchical Bayesian modelling is first introduced as a means of logically combining data from different sources to augment limited project-specific data. A Bayesian hierarchical non-linear regression model for the analysis of rock strength data is then developed and implemented; it is applied to 40 strength data sets of granite retrieved from the literature. In the context of these data, the advantages of the hierarchical model and the improvements in strength parameter estimations brought about by its application are discussed. Also discussed is the goodness-of-fit of the hierarchical model in comparison with more conventional statistical models. The paper concludes with suggestions for further development of the proposed hierarchical model, and the potential of hierarchical modelling as a general approach to statistical modelling of geotechnical data. [ABSTRACT FROM AUTHOR]

For centuries, the study of traumatic brain injury (TBI) has been centred on historical observation and analyses of personal, social, and environmental processes, which have been examined separately. Today, computation implementation and vast patient data repositories can enable a concurrent analysis of personal, social, and environmental processes, providing insight into changes in health status transitions over time. We applied computational and data visualization techniques to categorize decade-long health records of 235,003 patients with TBI in Canada, from preceding injury to the injury event itself. Our results highlighted that health status transition patterns in TBI emerged along with the projection of comorbidity where many disorders, social and environmental adversities preceding injury are reflected in external causes of injury and injury severity. The strongest associations between health status preceding TBI and health status at the injury event were between multiple body system pathology and advanced age-related brain pathology networks. The interwoven aspects of health status on a time continuum can influence post-injury trajectories and should be considered in TBI risk analysis to improve prevention, diagnosis, and care. [ABSTRACT FROM AUTHOR]

Background: The interplay of host, agent, and environment implicated in traumatic brain injury (TBI) events is difficult to account for in hypothesis-driven research. Data-driven analysis of injury data can enable insight into injury events in novel ways. This research dissected complex and multidimensional data at the time of the TBI event by exploiting data mining and information visualization methods.Methods: We drew upon population-based decade-long health administrative data collected through the routine operation of the publicly funded health system in Ontario, Canada. We applied a computational approach to categorize health records of 235,003 patients with TBI versus the same number of reference patients without TBI, individually matched based on sex, age, place of residence, and neighbourhood income quantile. We adopted the basic concepts of the Haddon Matrix (host, agent, environment) to organize emerging factors significantly related to TBI versus non-TBI events. To explore sex differences, the data of male and female patients with TBI were plotted on heatmaps and clustered using hierarchical clustering algorithms.Results: Based on detected similarities, the computational technique yielded 34 factors on which individual TBI-event codes were loaded, allowing observation of a set of definable patterns within the host, the agent, and the environment. Differences in the patterns of host, agent and environment were found between male and female patients with TBI, which are currently not identified based on data from injury surveillance databases. The results were internally validated.Conclusions: The study outlines novel areas for research relevant to TBI and offers insight into how computational and visual techniques can be applied to advance the understanding of TBI event. Results highlight unique aspects of sex differences of the host and agent at the injury event, as well as differences in exposure to adverse social and environmental circumstances, which can be a function of gender, aiding in future studies of injury prevention and gender-transformative care. [ABSTRACT FROM AUTHOR]

Objective: To understand how pre-injury health status present five-years preceding traumatic brain injury (TBI) affects direct medical cost two years post-injury. Methods: Patients age ≥19 years in the emergency department (ED) or acute care for a TBI between April 1, 2007 and March 31, 2014 in Ontario, Canada (N = 55,669) were identified from population-based health administrative data. Forty-three factors of pre-injury health status (i.e., comorbidities and personal, social, and environmental factors) that were internally validated for the TBI population were assessed in this study. The outcome of interest was direct medical cost within two years of discharge. Sex-specific multivariable linear regressions were conducted to understand the associations between direct medical cost within two years of discharge and pre-injury health status. Results: Patients who received care in the ED (81.9% of total sample) incurred a median cost of $2,492/male patient (average $12,342/patient) and $3,508/female patient (average $65,285/patient) within two years of injury; 37 pre-injury factors were significantly associated with increased direct medical costs. Patients who first received care for their TBI in acute care (18.1%) incurred a median cost of $25,081/male patient (average $63,060/patient) and $30,277/female patient (average $65,285/patient) within two years of injury; 21 factors were significantly associated with increased direct medical costs. Among more prevalent factors, those associated with increased medical cost by at least 50% included mental health disorders, substance abuse, disorders or medical conditions frequently observed among the elderly, cardiovascular disorders, stroke and emergencies involving the brain, metabolic disorders and abdominal symptoms, conditions and symptoms of abdomen and pelvis, genitourinary disorders and disorders of prostate, and pulmonary abdominal and other emergencies. Conclusions: Direct medical costs two years post-TBI differed significantly between patients with and without adverse pre-existing health status. Interdisciplinary teams to promote early identification of pre-existing health conditions and appropriate management and integration of these conditions in TBI care across the continuum of healthcare may be opportunities to reduce direct medical costs post-injury. [ABSTRACT FROM AUTHOR]

Focuses on a study which examined the prevalence of reading disability in boys and girls. Clinical characteristics of subjects; Criterion measures used; Characteristics of reading disability in the subjects.

Presents a response by Salley E. Shaywitz, Jack M. Fletcher and Michael D. Escobar to a letter to the editor about their article 'Prevalence of Reading Disability in Boys and Girls: Results of the Connecticut Longitudinal Study,' in the 1990 issue of the 'Journal of the American Medical Association.'