Bayesian kernel projections for classification of high dimensional data

A Bayesian multi-category kernel classification method is proposed. The algorithm performs the classification of the projections of the data to the principal axes of the feature space. The advantage of this approach is that the regression coefficients are identifiable and sparse, leading to large computational savings and improved classification performance. The degree of sparsity is regulated in a novel framework based on Bayesian decision theory. The Gibbs sampler is implemented to find the posterior distributions of the parameters, thus probability distributions of prediction can be obtained for new data points, which gives a more complete picture of classification. The algorithm is aimed at high dimensional data sets where the dimension of measurements exceeds the number of observations. The applications considered in this paper are microarray, image processing and near-infrared spectroscopy data.     

An iterative Monte Carlo method for nonconjugate Bayesian analysis

The Gibbs sampler has been proposed as a general method for Bayesian calculation in Gelfand and Smith (1990). However, the predominance of experience to date resides in applications assuming conjugacy where implementation is reasonably straightforward. This paper describes a tailored approximate rejection method approach for implementation of the Gibbs sampler when nonconjugate structure is present. Several challenging applications are presented for illustration.     

AN EXACT CONDITIONAL TEST FOR COVARIANCE SELECTION MODELS

An exact conditional test is developed for testing the absence of an edge in a graphical covariance selection model and is shown to be equivalent to a test based on the partial correlation coefficient. An example is given.     

Reclassifying causes of death to study the epidemiological transition in The Netherlands, 1875–1992

This article describes a method for reclassifying causes of death in the Netherlands for the period 1875–1992. Two criteria should be met to obtain a useful classification to study the epidemiological transition. First, the categories should be nosologically continuous over the period under study and second, there should be enough detail in causes which are important in the context of the epidemiological transition viz. communicable diseases, non-communicable diseases and external causes of death. A method developed by Vallin and Meslé (1988), which involves ‘dual correspondence tables’ and ‘fundamental associations’, was used to create nosologically continuous categories. These categories were tested for statistical continuity during the transition years of one ICD-revision to the next, using ordinary least squares regression analysis. The reclassification procedure resulted in a nested classification consisting of three levels of refinement of causes of death: 27 causes, 1875–1992; 65 causes, 1901–1992; and 92 causes, 1931–1992. On the basis of this classification, 43% of all deaths in 1875–79 and 98% of all deaths in 1992 could be allocated to either communicable diseases, non-communicable diseases or external causes.     

On simultaneous closeness probabilities of order statistics from odd sample sizes to the population median

In this note, we present alternative derivations for the probability that an individual order statistic is closest to the target parameter among all order statistics from a complete random sample. This approach is simpler than the geometric arguments used earlier. We also provide a simple direct proof for the symmetry property of the simultaneous closeness probabilities among order statistics for the estimation of percentiles from a symmetric family. Finally, we offer an alternative simpler proof for the result that sample medians from larger odd sample sizes are Pitman closer to the population median than sample medians from smaller odd sample sizes.     

On Integer Programming Approaches for Morphological Template Decomposition Problems in Computer Vision

In morphological image processing and analysis, a template or structuringelement is applied to an image. Often savings in computation time and abetter fit to the given computer architecture can be achieved by using thetechnique of template decomposition. Researchers have written a multitude ofpapers on finding such decompositions for special classes of templates.Justifying recent integer programming approaches to the morphologicaltemplate decomposition problem in its general form, this paper proves theNP-completeness of this problem.     

The role of public health in population health

The role of public health is central in population health. What distinguishes public health from clinical medicine is that it's focus is on the entire population--not the individual patient. To achieve the goal of population health, healthy communities, or Healthy People 2000 requires physician leadership in medicine and public health at all levels--local, state, and national. The challenge is formidable, but the goals are attainable through strategies that focus on the goal of the common good and through collaboration of public health, medicine, and the community.     

Meet the mandate to measure clinical quality

The mandate for health care organizations to be accountable for quality, as well as price, is now unavoidable. The Joint Commission's ORYX project is requiring every hospital to measure clinical outcomes of a majority of its patients within the next three years. This mandate can be met best with systems of clinical outcomes measurement that provide valid, reliable risk adjustment; yield meaningful information about many different diseases and procedures; and measure more than mortality or cost--all using primarily billing data. New outcomes measurement tools with all of these capabilities are available and have already enabled quality improvement in dozens of hospitals across the U.S.     

Changes in physician and non-physician relationships

At the end of World War II, one-third of the nation's hospital administrators were physicians. During the 1950's through the mid-1980's a new breed of masters'level administrator, with well-honed coordinating skills, orchestrated a major expansion of new programs, services, and facilities. With the advent of the Medicare prospective payment system (PPS), more governing boards restructured their administrative staffs with corporate titles. Meanwhile, physicians sensed that trustees were becoming far more concerned with bottom line performance to repay a mounting debt that hospitals had incurred to remain technologically competitive. Since mergers and integrated health systems by themselves will be unable to generate significant operating efficiencies, governing boards will be forced to change direction and shift back to recruiting physicians as their CEOs or in other senior positions to assure themselves of the clinical leadership required to implement the managed care concepts of reducing utilization and cost, and simultaneously enhancing quality of patient care.     

ATSDR Science Panel on the Bioavailability of Mercury in Soils: Lessons Learned

On the basis of discussion and analysis during and following an ATSDR science panel on the bioavailability of mercury in soils, it is apparent that the default assumption of 100% relative bioavailability for mercury-contaminated soils is excessively conservative. However, current knowledge does not allow the development of default assumptions or guidelines for determining relative bioavailability of mercury in soils. Until such default assumptions or guidelines can be developed, site-specific assays of bioavailability, preferably using either animal bioassays or validated in vitro techniques, may provide the best approach for estimating soil-mercury bioavailability.