Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “model selection”

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 451 records · Page 25Linked to original sources
PubMed2004

New results on error correcting output codes of kernel machines.

We study the problem of multiclass classification within the framework of error correcting output codes (ECOC) using margin-based binary classifiers. Specifically, we address two important open problems in this context: decoding and model selection. The decoding problem concerns how to map the outputs of the classifiers into class codewords. In this paper we introduce a new decoding function that combines the margins through an estimate of their class conditional probabilities. Concerning model selection, we present new theoretical results bounding the leave-one-out (LOO) error of ECOC of kernel machines, which can be used to tune kernel hyperparameters. We report experiments using support vector machines as the base binary classifiers, showing the advantage of the proposed decoding function over other functions of the margin commonly used in practice. Moreover, our empirical evaluations on model selection indicate that the bound leads to good estimates of kernel parameters.

Neural Networks, Computer↗
PubMed1983

Density dependent selection incorporating intraspecific competition 1. A haploid model.

A haploid model is introduced and analyzed in which intraspecific competition is incorporated within a density dependent framework. It is assumed that each genotype has a unique carrying capacity corresponding to the equilibrium population size when fixed for that type. Each genotypic fitness at a single multi-allelic locus is a function of a distinctive effective population size formed by adding the numbers of each genotype present, weighted by an intraspecific competition coefficient. As a result, the fitnesses depend upon the relative frequencies of the various genotypes as well as the total population size. Intergenotypic interactions can have a profound effect upon the outcome of the population. In particular, when the density effect of one individual upon another depends upon their respective genotypes, a unique stable interior equilibrium is possible in which all alleles are present. This stands in contrast to the purely density dependent haploid system in which the only possible stable state corresponds to fixation for the type with the highest carrying capacity. In the present model selective advantage is determined by a balance between carrying capacity and sensitivity to density pressures from other genotypes. Fixation for the genotype with the highest carrying capacity, for instance, will not be stable if it exerts a sufficiently weak competitive effect upon the other genotypes. In the diallelic case, maintenance of both alleles at a stable equilibrium requires that the net intragenotypic competition between individuals of like genotype be stronger than that between unlike types. As for purely density regulated systems, there may be no stable equilibria and/or regular and chaotic cycling may occur. The results may also be interpreted in terms of a discrete time model of interspecific competition with each haplotype representing a different species.

Animals↗
PubMed2004

Statistical modeling for selecting housekeeper genes.

There is a need for statistical methods to identify genes that have minimal variation in expression across a variety of experimental conditions. These 'housekeeper' genes are widely employed as controls for quantification of test genes using gel analysis and real-time RT-PCR. Using real-time quantitative RT-PCR, we analyzed 80 primary breast tumors for variation in expression of six putative housekeeper genes (MRPL19 (mitochondrial ribosomal protein L19), PSMC4 (proteasome (prosome, macropain) 26S subunit, ATPase, 4), SF3A1 (splicing factor 3a, subunit 1, 120 kDa), PUM1 (pumilio homolog 1 (Drosophila)), ACTB (actin, beta) and GAPD (glyceraldehyde-3-phosphate dehydrogenase)). We present appropriate models for selecting the best housekeepers to normalize quantitative data within a given tissue type (for example, breast cancer) and across different types of tissue samples.

ATPases Associated with Diverse Cellular Activitie↗
PubMed1997

Alpha 1-adrenoceptor subtype selectivity: molecular modelling and theoretical quantitative structure-affinity relationships.

This study constitutes a preliminary rationalization, at the molecular level, of antagonist selectivity towards the three cloned alpha 1-adrenergic receptor (alpha 1-AR) subtypes. Molecular dynamics simulations allowed a structural/dynamics analysis of the seven alpha-helix-bundle models of the bovine alpha 1a-, hamster alpha 1b-, and rat alpha 1d-AR subtypes. The results showed that the transmembrane domains of these subtypes have different dynamic behaviours and different topographies of the binding sites, which are mainly constituted by conserved residues. In particular, the alpha 1a-AR binding site is more flexible and topographically different with respect to the other two subtypes. The results of the theoretical structural/dynamics analysis of the isolated receptors are consistent with the binding affinities of the 16 antagonists tested towards the three cloned alpha 1-AR subtypes. Moreover, the theoretical quantitative structure-affinity relationships obtained from the antagonist-receptor interaction models further corroborates the hypothesis that selectivity towards one preferential subtype is mainly modulated by receptor and/or ligand distortion energies. In other words, subtype selectivity seems to be mainly guided by the dynamic complementarity (induced fit) between ligand and receptor. On the basis of the quantitative models presented it is possible to predict both affinities and selectivities of putative alpha 1-AR ligands as well as to estimate the theoretical alpha 1-AR subtype affinities and selectivities of existing antagonists.

Adrenergic alpha-1 Receptor Antagonists↗
PubMed1989

Likelihood estimation of quantitative genetic parameters when selection occurs: models and problems.

Conceptual aspects of estimation of genetic components of variance and covariance under selection are discussed, with special attention to likelihood methods. Certain selection processes are described and alternative likelihoods that can be used for analysis are specified. There is a mathematical relationship between the likelihoods that permits comparing the relative amount of information contained in them. Theoretical arguments and evidence indicate that point inferences made from likelihood functions are not affected by some forms of selection.

Animals↗
PubMed1990

Quality and provider choice: a multinomial logit-least-squares model with selectivity.

A Federal Trade Commission survey of contact lens wearers is used to estimate a multinomial logit-least-squares model of the joint determination of provider choice and quality of care in the contact lens industry. The effect of personal and industry characteristics on a consumer's choice among three types of providers--opticians, ophthalmologists, and optometrists--is estimated via multinomial logit. The regression model of the quality of care has two features that distinguish it from previous work in the area. First, it uses an outcome rather than a structural or process measure of quality. Quality is measured as an index of the presence of seven potentially pathological eye conditions caused by poorly fitted lenses. Second, the model controls for possible selection bias that may arise from the fact that the sample observations on quality are generated by consumers' nonrandom choices of providers. The multinomial logit estimates of provider choice indicate that professional regulations limiting the commercial practices of optometrists shift demand for contact lens services away from optometrists toward ophthalmologists. Further, consumers are more likely to have their lenses fitted by opticians in states that require the licensing of opticians. The regression analysis of variations in quality across provider types shows a strong positive selection bias in the estimate of the quality of care received by consumers of ophthalmologists' services. Failure to control for this selection bias results in an overestimate of the quality of care provided by ophthalmologists.

Adolescent↗
PubMed2002

The effect of precision of molecular orbital descriptors on toxicity modeling of selected pyridines.

The response-surface approach to QSARs attempts to model toxic potency of diverse groups of chemicals while avoiding problems associated with the identification of the mechanism of toxic action or specific chemical class often associated with other approaches. However, while hydrophobicity-dependent, simple regression QSARs derived for congeneric series of organic compounds typically have coefficients of determination greater than 0.90, more heterogeneous multiple regression QSARs exhibit typically 10-15% more unexplained variability. One difference between these approaches is the use of a quantum chemical (QC) descriptor, particularly molecular orbital (MO) energy values such as the energy of the lowest unoccupied molecular orbital (E(LUMO)). The reduced statistical fit exhibited by QSAR models, which include these QC-MO descriptors, could be a result of the variability inherent in the calculation of these descriptors. The present investigation with a structurally and mechanistically diverse set of pyridines revealed that variability is associated with the calculation of the MO descriptor E(LUMO) both between selected Hamiltonians and selected software packages. However, this variability in no way affects the statistical significance of QSARs for toxicity using these values. Specifically, the E(LUMO) values calculated with the PM3 and AM1 Hamiltonians in the two software packages were highly related. There was no relationship between molecular complexity or chemical reactivity and increased differences in individual ELUMO values as described by the standard errors of the mean. Although nine appeared to be the number of calculations, which best minimizes the standard error in energy values relative to computational costs; this minimization did not alter the statistics of the QSARs derived with single vs. mean E(LUMO) values. While the energy of the highest occupied molecular orbital (E(HOMO)) values were not used in the modeling of toxicity, a comparison of these values revealed greater variability between the Hamiltonians and software packages than observed for ELUMO values. Examination of the magnitudes of standard error of the E(HOMO) values in connection to structural features or reactivity likewise revealed no trends.

Forecasting↗
PubMed2000

Detection of bias in harvest-based estimates of chronic wasting disease prevalence in mule deer.

Diseased animals may exhibit behavioral shifts that increase or decrease their probability of being randomly sampled. In harvest-based sampling approaches, animal movements, changes in habitat utilization, changes in breeding behaviors during harvest periods, or differential susceptibility to harvest via behaviors like hiding or decreased sensitivity to stimuli may result in a non-random sample that biases prevalence estimates. We present a method that can be used to determine whether bias exists in prevalence estimates from harvest samples. Using data from harvested mule deer (Odocoileus hemionus) sampled in northcentral Colorado (USA) during fall hunting seasons 1996-98 and Akaike's information criterion (AIC) model selection, we detected within-yr trends indicating potential bias in harvest-based prevalence estimates for chronic wasting disease (CWD). The proportion of CWD-positive deer harvested slightly increased through time within a yr. We speculate that differential susceptibility to harvest or breeding season movements may explain the positive trend in proportion of CWD-positive deer harvested during fall hunting seasons. Detection of bias may provide information about temporal patterns of a disease, suggest biological hypotheses that could further understanding of a disease, or provide wildlife managers with information about when diseased animals are more or less likely to be harvested. Although AIC model selection can be useful for detecting bias in data, it has limited utility in determining underlying causes of bias. In cases where bias is detected in data using such model selection methods, then design-based methods (i.e., experimental manipulation) may be necessary to assign causality.

Age Factors↗
PubMed2005

Risk factors of prolonged postoperative pleural effusion after repair of tetralogy of Fallot.

BACKGROUND: Tetralogy of Fallot (TOF) is the most common cyanotic congenital heart disease, and total correction is the definitive treatment. Chest tube drainage of pleural effusion (PE) is essential after surgery. Prolonged PE (> 7 days) is one of the complications; it may increase hospital stay and the risks of morbidity and mortality. The aim of this study was to investigate and analyze the possible risk factors for prolonged PE after total correction of TOF. METHODS: Thirty-seven patients who received total correction of TOF between July 1999 and April 2001 were included in this study. They were divided into 2 groups according to the duration of chest tube drainage for postoperative PE: Group I had postoperative PE < or = 7 days; Group II had postoperative PE > 7 days. Detailed records were taken on patients' demographic characteristics, blood parameters, surgery, electrocardiographic and radiologic data, and angiographic and echocardiographic findings. The data of the 2 groups were compared using the Wilcoxon rank-sum test and Fisher's exact test. Risk factors were analyzed by logistic regression and model selection. RESULTS: Of the 37 patients, 16 were male and 21 were female. There were 32 patients (86.5%) in Group I and 5 (13.5%) in Group II. Mean patient age at repair was 1.82 +/- 1.29 years (range, 0.53-3.11 years). Significant differences (p < 0.05) between the 2 groups were noted for gender, age at repair, body weight, presence of wound infection, duration on heart-lung machine (bypass time), oxygen saturation before surgery, duration of endotracheal intubation, length of hospital stay, and Nakata index. These risk factors were analyzed by logistic regression and model selection. Two models were set up: Model 1--oxygen saturation before surgery, presence of wound infection, age at repair; Model 2--oxygen saturation before surgery, presence of wound infection. CONCLUSION: Prolonged PE is a significant morbidity after TOF repair. The risk factors for prolonged PE are gender, age at repair, body weight, bypass time, low oxygen saturation before surgery, wound infection after surgery, duration of endotracheal intubation, length of hospital stay, and Nakata index. Oxygen saturation before surgery and wound infection were major risk factors while age at repair was a confounder.

Age Factors↗
PubMed2005

[Selection of animal model in myopia research].

Animal model plays an important role in myopia research. However discrepancy and deputed have been existed in animal model making, including animal selection, myopia inducing methods. The model selection can not only determine the success of experiments but also impact the accuracy of study outcome. Here are summaries and comments round disputed issues which are hoped to be useful for animal myopia researches.

Animals↗
PubMed2006

Evaluating mortality in intensive care units: contribution of competing risks analyses.

INTRODUCTION: Kaplan-Meier curves and logistic models are widely used to describe and explain the variability of survival in intensive care unit (ICU) patients. The Kaplan-Meier approach considers that patients discharged alive from hospital are 'non-informatively' censored (for instance, representative of all other individuals who have survived to that time but are still in hospital); this is probably wrong. Logistic models are adapted to this so-called 'competing risks' setting but fail to take into account censoring and differences in exposure time. To address these issues, we exemplified the usefulness of standard competing risks methods; namely, cumulative incidence function (CIF) curves and the Fine and Gray model. METHODS: We studied 203 mechanically ventilated cancer patients with acute respiratory failure consecutively admitted over a five-year period to a teaching hospital medical ICU. Among these patients, 97 died before hospital discharge. After estimating the CIF of hospital death, we used Fine and Gray models and logistic models to explain variability hospital mortality. RESULTS: The CIF of hospital death was 35.5% on day 14 and was 47.8% on day 60 (97/203); there were no further deaths. Univariate models, either the Fine and Gray model or the logistic model, selected the same eight variables as carrying independent information on hospital mortality at the 5% level. Results of multivariate were close, with four variables selected by both models: autologous stem cell transplantation, absence of congestive heart failure, neurological impairment, and acute respiratory distress syndrome. Two additional variables, clinically documented pneumonia and the logistic organ dysfunction, were selected by the Fine and Gray model. CONCLUSION: The Fine and Gray model appears of interest when predicting mortality in ICU patients. It is closely related to the logistic model, through direct modeling of times to death, and can be easily extended to model non-fatal outcomes.

Adult↗
PubMed1998

Visualization of CD4/CD8 T cell commitment.

A system to innocuously visualize T cell lineage commitment is described. Using a "knock-in" approach, we have generated mice expressing a beta-galactosidase reporter in place of CD4; expression of beta-galactosidase in these animals appears to be an accurate and early indicator of CD4 gene transcription. We have exploited this knock-in line to trace CD4/CD8 lineage commitment in the thymus, avoiding important pitfalls of past experimental approaches. Our results argue in favor of a selective model of thymocyte commitment, demonstrating a fundamentally symmetrical process: engagement of either class of major histocompatibility complex (MHC) molecule by a differentiating CD4(+)CD8(+) cell can give rise to T cell antigen receptor (TCR)hi thymocytes of either lineage. Key findings include (a) direct demonstration of a substantial number of CD4-committed, receptor/coreceptor-mismatched cells in MHC class II- deficient mice, a critical prediction of the selective model; (b) highly efficient rescue of such "mismatched" intermediates by forced expression of CD8 in a TCR transgenic line, and an explanation of why previous experiments of this nature were less successful-a major past criticism of the selective model; (c) direct demonstration of an analogous, though smaller, population of CD8-committed mismatched intermediates in class I-deficient animals. Finally, we found no evidence of a CD4 default pathway.

Animals↗
PubMed1990

The molecular nature of allelic diversity for two models of balancing selection.

This paper provides a theoretical description of the distribution of the number of mutations that separate alleles that are held in a population by balancing selection. Two models of nucleotide site epistasis are described: parity models and additive site models. Parity models are shown to result in a more uniform distribution of mutations across alleles than the neutral model, while additive sites models show a more extreme distribution. The analytic approach uses strong-selection, weak-mutation approximations to constant-fitness and random-environment diffusion models.

Base Sequence↗
PubMed1995

Formation of membrane domains created during the budding of vesicular stomatitis virus. A model for selective lipid and protein sorting in biological membranes.

Vesicular stomatitis virus buds from domains of the plasma membrane that have a unique protein and lipid composition. Fluorescence digital imaging microscopy and resonance energy transfer were used to determine how the two viral envelope-associated proteins, the G and the M proteins, could alter the lateral distribution of lipids in large unilamellar vesicles and form domains. The G protein formed large domains in vesicles containing phosphatidic acid but not with phosphatidylserine, while the M protein formed domains enriched in both acidic phospholipids. Domains enriched in sphingomyelin were observed only when both the G protein and the M protein were present in vesicles containing phosphatidic acid. Phosphatidylcholine and gramicidin (chosen to represent a host membrane protein) were excluded from the domains. Cholesterol was induced to partition into the domains only in vesicles containing phosphatidic acid and sphingomyelin along with both of the proteins. Phosphatidylethanolamine was not enriched or depleted in the domains. Domains of similar composition were formed using vesicles made from dioleoylphospholipids and the lipids extracted from BHK-21 plasma membranes, indicating that the fatty acid composition was not as important as the polar head groups of the phospholipids. The phospholipid and cholesterol compositions of the domains formed by the G and the M proteins in vesicles were very similar to the composition of the viral envelope, suggesting that the domains represent the areas in the plasma membrane where the virus buds. This study provides a model for selective lipid and protein sorting that occurs in biological membranes.

Animals↗
PubMed1993

A simple economic evaluation model for selecting diabetes health care strategies.

The aim of this work was to develop a simple model to evaluate the health care cost of diabetes in order to help decision-makers in selecting strategies for the prevention, control, and treatment of the disease. The model assesses the cost of full coverage of health care of diabetic patients free of chronic complications and in comparison with those of the treatment of acute and chronic complications. For that purpose, standardized cost figures were obtained from both annual control and treatment of uncomplicated diabetic patients and the treatment of one episode of ketoacidosis, acute myocardial infarction, and amputation of two toes. The treatment/prevention cost ratio obtained showed that prevention of a single episode of any of these late-complication-related events would provide enough funds to cover either the total or partial annual cost of control and treatment of several uncomplicated diabetic patients. These facts would favour the allocation of funds in such a way as to allow adequate control and treatment of diabetic patients to keep them free of the chronic complications of the disease. This would be a wise investment of funds, which would result in a reduction in the socioeconomic cost of the disease and in a better quality of life for the diabetic patients and their families.

Argentina↗
PubMed1991

Microscopic model for selective permeation in ion channels.

Ionic permeation in the selectivity filter of ion channels is analyzed by a microscopic model based on molecular kinetic theory. The energy and flux equations are derived by assuming that: (a) the selectivity filter is formed by a symmetrical array of carbonyl groups; (b) ion movement is near the axis of the channel; (c) a fraction of water molecules is separated from the ion while it moves across the selectivity filter; (d) the applied voltage drops linearly across the selectivity filter; (e) ions move independently. Energy profiles, single channel conductances, and the degree of hydration of K+ in a hypothetical K+ channel are examined by varying the following microscopic parameters: ion radius and mass, channel radius, number of effective water dipoles, and number of carbonyl groups. The i-V curve is linear up to +/- 170 mV. If the positions of energy maxima and minima are fixed, this linear range is reduced to +/- 50 mV. Channel radius and ion-water interactions are found to be two major channel structural determinants for selectivity sequences. Both radius and mass of an ion are important in selectivity mediated by these interactions. The theory predicts a total of 15 possible kinetic selectivity sequences for alkali cations in ion channels with a single selectivity filter.

Animals↗