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 955 records · Page 53Linked to original sources
PubMed2005

Perfusion-diffusion compartmental models describe cerebral helium kinetics at high and low cerebral blood flows in sheep.

This study evaluated the relative importance of perfusion and diffusion mechanisms in compartmental models of blood:tissue helium exchange in the brain. Helium has different physiochemical properties from previously studied gases, and is a common diluent gas in underwater diving where decompression schedules are based on theoretical models of inert gas kinetics. Helium kinetics across the cerebrum were determined during and after 15 min of helium inhalation, at separate low and high steady states of cerebral blood flow in seven sheep under isoflurane anaesthesia. Helium concentrations in arterial and sagittal sinus venous blood were determined using gas chromatographic analysis, and sagittal sinus blood flow was monitored continuously. Parameters and model selection criteria of various perfusion-limited or perfusion-diffusion compartmental models of the brain were estimated by simultaneous fitting of the models to the sagittal sinus helium concentrations for both blood flow states. Purely perfusion-limited models fitted the data poorly. Models that allowed a diffusion-limited exchange of helium between a perfusion-limited tissue compartment and an unperfused deep compartment provided better overall fit of the data and credible parameter estimates. Fit to the data was also improved by allowing countercurrent diffusion shunt of helium between arterial and venous blood. These results suggest a role of diffusion in blood:tissue helium equilibration in brain.

Animals↗
PubMed2002

Contrasting models for lactation curve analysis.

Several statistical models have been proposed for the genetic evaluation of production traits in dairy cattle based on test-day records. Three main approaches have been put forward in the literature: random regression, orthogonal polynomials, and, more recently, character process models. The aim of this paper is to show how these different approaches are related, to compare their performance for the genetic analysis of lactation curves, and to assess equivalence between sire and animal models for repeated measures analyses. It was found that, with an animal model, a character process model with 11 parameters performed better, regarding the likelihood criterion, than a quartic random regression model (with 31 parameters). However, although the likelihood was higher, the genetic variance was very different with the character process model from the unstructured model, which raises important issues concerning model selection criteria. There are advantages in combining methodologies. A quadratic random regression model for the environmental part, combined with a character process model for the residual, performed better than the quartic random regression model and had fewer parameters. A character process structure allowing for a correlation pattern modeled the residual better than a simple quadratic variance, and had only one extra parameter.

Animals↗
PubMed2006

Wild bird mortality and West Nile virus surveillance: biases associated with detection, reporting, and carcass persistence.

Surveillance targeting dead wild birds, in particular American crows (Corvus brachyrhynchos), plays a critical role in West Nile virus (WNV) surveillance in the United States. Using crow decoy surrogates, detection and reporting of crow carcasses within urban and rural environments of DeKalb County, Georgia were assessed for potential biases that might occur in the county's WNV surveillance program. In each of two replicated trials, during July and September 2003, 400 decoys were labeled with reporting instructions and distributed along randomly chosen routes throughout designated urban and rural areas within DeKalb County. Information-theoretic methods were used to compare alternative models incorporating the effects of area and trial on probabilities of detection and reporting. The model with the best empirical support included the effects of area on both detection and reporting of decoys. The proportion of decoys detected in the urban area (0.605, SE=0.024) was approximately twice that of the rural area (0.293, SE=0.023), and the proportion of decoys reported in the urban area (0.273, SE=0.023) was approximately three times that of the rural area (0.103, SE=0.028). These results suggest that human density and associated factors can substantially influence dead crow detection and reporting and, thus, the perceived distribution of WNV. In a second and separate study, the persistence and fate of American crow and house sparrow (Passer domesticus) carcasses were assessed in urban and rural environments in Athens-Clarke, Madison, and Oconee counties, Georgia. Two replicated trials using 96 carcasses of each species were conducted during July and September 2004. For a portion of the carcasses, motion sensitive cameras were used to monitor scavenging species visits. Most carcasses (82%) disappeared or were decayed by the end of the 6-day study. Carcass persistence averaged 1.6 days in rural areas and 2.1 days in urban areas. We analyzed carcass persistence rates using a known-fate model framework in program MARK. Model selection based on Akaike's Information Criteria (AIC) indicated that the best model explaining carcass persistence rates included species and number of days of exposure; however, the model including area and number of days of exposure received approximately equal support. Model-averaged carcass persistence rates were higher for urban areas and for crow carcasses. Six mammalian and one avian species were documented scavenging upon carcasses. Dead wild birds could represent potential sources of oral WNV exposure to these scavenging species. Species composition of the scavenger assemblage was similar in urban and rural areas but "scavenging pressure" was greater in rural areas.

Animals↗
PubMed1999

Multilevel modeling and model averaging.

Multilevel modeling, also known as hierarchical regression, generalizes ordinary regression modeling to allow explicit and flexible compromises between simple and complex models. This article provides an elementary introduction to multilevel modeling as a model-averaging technique. Model averaging provides an alternative to model selection, and it emphasizes the role of prior information in finding good models.

Abortion, Spontaneous↗
PubMed1990

What is natural fertility? The remodeling of a concept.

This paper applies three log-linear models to Louis Henry's original 1961 natural fertility data in order to test various assumptions leading to ways of obtaining a standard natural fertility schedule through explicit modeling. "The models specify that births follow an independent Poisson distribution for each age interval of each population. All parameters are estimated through an iterative maximum-likelihood procedure." The author suggests that the model selected provides better estimates of the standard natural fertility function than previous models.

Demography↗
PubMed1987

Biochemical and pharmacologic rationale for high-dose methotrexate.

High-dose methotrexate (HDMTX) regimens were initially designed to overcome methotrexate (MTX) resistance due to defective transport of the drug. At high concentrations, enough MTX diffuses into resistant cells to saturate and inhibit the target enzyme, dihydrofolate reductase (DHFR). The high doses of MTX needed to achieve these high drug concentrations must be administered with the reduced folate antidote, leucovorin (LV; 5-formyltetrahydrofolate), to prevent increased toxicity. To increase MTX therapeutic index, LV rescue must be selective, i.e., more effective in normal than in tumor cells. In experimental models, selective rescue can be achieved if strict LV administration guidelines are respected. Since both MTX and LV use the membrane transport system, it was hypothesized that selective rescue occurred because transport-deficient, MTX-resistant tumor cells also transported LV poorly. The unsatisfactory clinical results frequently obtained with HDMTX regimens suggest a need to re-evaluate this underlying rationale, especially in view of recent findings concerning the mechanisms of MTX resistance and LV rescue. Experimentally, cells resistant to MTX because of an increased or altered DHFR, decreased metabolism to polyglutamates, or decreased thymidylate synthase activity are not always significantly more sensitive to higher concentrations of MTX. Furthermore, recent studies on human small cell lung cancer cell lines suggest that decreased MTX polyglutamate metabolism and thymidylate synthase activity might be prevalent MTX-resistant mechanisms in human tumors. Selective LV rescue could also occur through mechanisms other than selective uptake by normal tissues.(ABSTRACT TRUNCATED AT 250 WORDS)

Biological Transport↗
PubMed1983

Path analysis in genetic epidemiology: a critique.

Path analysis, a form of general linear structural equation models, is used in studies of human genetics data to discern genetic, environmental, and cultural factors contributing to familial resemblance. It postulates a set of linear and additive parametric relationships between phenotypes and genetic and cultural variables and then essentially uses the assumption of multivariate normality to estimate and perform tests of hypothesis on parameters. Such an approach has been advocated for the analysis of genetic epidemiological data by D. C. Rao, N. Morton, C. R. Cloninger, L. J. Eaves, and W. E. Nance, among others. This paper reviews and evaluates the formulations, assumptions, methodological procedures, interpretations, and applications of path analysis. To give perspective, we begin with a discussion of path analysis as it occurs in the form of general linear causal models in several disciplines of the social sciences. Several specific path analysis models applied to lipoprotein concentrations, IQ, and twin data are then reviewed to keep the presentation self-contained. The bulk of the critical discussion that follows is directed toward the following four facets of path analysis: (1) coherence of model specification and applicability to data; (2) plausibility of modeling assumptions; (3) interpretability and utility of the model; and (4) validity of statistical and computational procedures. In the concluding section, a brief discussion of the problem of appropriate model selection is presented, followed by a number of suggestions of essentially model-free alternative methods of use in the treatment of complex structured data such as occurs in genetic epidemiology.

Asian↗
PubMed1996

Relationship between the humidity and temperature of inspired gas and the function of the airway mucosa.

OBJECTIVE: To review the available literature on the relationship between the humidity and temperature of inspired gas and airway mucosal function. DATA SOURCES: International computerized databases and published indices, experts in the field, conference proceedings, bibliographies. STUDY SELECTION/DATA EXTRACTION: Two hundred articles/texts on respiratory tract physiology and humidification were reviewed. Seventeen articles were selected from 40 articles for inclusion in the published data verification of the model. Selection was by independent reviewers. Extraction was by consensus, and was based on finding sufficient data. DATA SYNTHESIS: A relationship exists between inspired gas humidity and temperature, exposure time to a given humidity level, and mucosal function. This relationship can be modeled and represented as an inspired humidity magnitude vs. exposure time map. The model is predictive of mucosal function and can be partially verified by the available literature. It predicts that if inspired humidity deviates from an optimal level, a progressive mucosal dysfunction begins. The greater the humidity deviation, the faster the mucosal dysfunction progresses. CONCLUSIONS: A model for the relationship between airway mucosal dysfunction and the combination of the humidity of inspired gas and the duration over which the airway mucosa is exposed to that humidity is proposed. This model suggests that there is an optimal temperature and humidity above which, and below which, there is impaired mucosal function. This optimal level of temperature and humidity is core temperature and 100% relative humidity. However, existing data are only sufficient to test this model for gas conditions below core temperature and 100% relative humidity. These data concur with the model in that region. No studies have yet looked at this relationship beyond 24 hrs. Longer exposure times to any given level of inspired humidity and inspired gas temperatures and humidities above core temperature and 100% relative humidity need to be studied to fully verify the proposed model.

Animals↗
PubMed1992

Genetic and environmental factors in the immune pathogenesis of rheumatoid arthritis.

Our experiments have led us to conclude that the rheumatoid arthritis shared epitope may act as a peptide that is important for positive and negative selection of T lymphocytes, that T lymphocytes are skewed by positive selection to recognize epitopes that are similar but not identical to self, and that peptide sequences that are similar to the RA-shared epitope are abundantly expressed by microorganisms that chronically infect most people. This combination of events could partly explain the association of the shared epitope with the severe forms of RA. The hypothesis cannot be tested directly, because we do not postulate that any unique population of autoreactive T cells is expanded in RA; however, the role of positive selection in molding the human T-cell repertoire to exogenous antigens can be tested by mapping T-cell antigenic determinants on the E. coli dnaJ protein or the gp110 protein of EBV in people with different HLA-DR types. Moreover, positive selection models imply that maternal antigens that cross the placenta can influence the T-cell repertoire. Thus, one might expect to find that the frequency of HLA-DR4 in the mothers of patients with RA who themselves lack the DR4 antigen, would be more frequent than predicted by chance alone. As the principles of positive selection are more precisely delineated in animal systems, it should become possible to ascertain more clearly how the shared epitope on HLA-DR molecules enhances the severity of autoimmune reactions; however, RA only occurs in humans; possibly because of the unique inability of human macrophages to replicate. Thus, only the direct analysis of patients can directly reveal the mechanisms of disease pathogenesis.

Arthritis, Rheumatoid↗
PubMed1994

The effect of scale dependent processes on kin selection: mating and density regulation.

Kin selection models describe fitness determining interactions that occur within small clusters of individuals often referred to as trait groups (Wilson, 1975). Because each individual influences not only its own fitness, but also the fitness of other members of the trait group, kin selection has both inter- and intra-group components (Hamilton, 1975; Wade, 1980). The importance of intergroup selection depends on the magnitude of inter-group differences in fertility and on the amount of genetic variation among trait groups. Migration and density regulation are likely to be important determinants of productivity and variation. Existing models integrate density regulation and migration structure into kin selection theory by applying carrying capacities or migration rates to trait groups. This requires that dispersal and density regulation both operate at the spatial scale of trait groups and excludes many cases of biological interest. I present a model that allows interaction, density regulation, and dispersion to operate over distinct spatial scales. The results of existing models can be retrieved as special cases of the general framework developed here. This model indicates that the appropriate scale for studying kin selection is determined by the spatial area over which the population is regulated. In effect, only a fraction of the total inter-trait group variance is available to intergroup selection. Processes that generate genetic variation over spatial scales larger than the "regulation scale" cannot aid the evolution of altruism. The analysis of a specific mating/migration model indicates how the various components of variation are determined by biological parameters such as migration rate. These results have important implications for the study of relatedness in natural populations.

Animals↗
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↗
PubMed1989

Anthelmintic resistance and the control of ovine ostertagiasis: a drug action model for genetic selection.

A site-specific genetic prediction model is presented, which examines the influences of different anthelmintic treatment regimens on selecting for drug resistance within a sheep management system. The model exploits the power of modern microcomputers to integrate factors such as parasite strain, geographic location, management practice and genetic fitness to identify effective control regimens which do not lead to resistant Ostertagia circumcincta strains over a 30-year horizon. The potential use of the model as a farm level management-support tool is discussed.

Animals↗
PubMedDate not supplied

Models for natural killer cell repertoire formation.

Natural killer (NK) cells lyse only cells that do not express sufficient levels of self class I MHC molecules. Inhibition of lysis is mediated by inhibitory receptors expressed by NK cells, such as the murine Ly49 receptors, that bind to MHC class I molecules. Since inhibitory receptor genes and MHC class I genes are located on different chromosomes, and are hence not automatically co-inherited, NK cells apparently adapt to the MHC environment during their development. Two models have been proposed to account for this "education" process of NK cells. The two-step selection model postulates that developing NK cells initiate the stable expression of a random set of Ly49 genes, and then undergo two selection steps, one for cells that express a sufficient number of self-MHC receptors, and one against cells that express too many inhibitory receptors. The sequential model postulates that a cell keeps initiating the stable expression of additional inhibitory receptors until a sufficient expression level of self-MHC specific receptors is reached, and the cell matures. In this study we implement both models in computer simulations, and compare simulation results to experimental data, in order to evaluate the relative plausibility of the two models.

Animals↗
PubMed2006

A copula model for repeated measurements with non-ignorable non-monotone missing outcome.

A normal copula-based selection model is proposed for continuous longitudinal data with a non-ignorable non-monotone missing-data process. The normal copula is used to combine the distribution of the outcome of interest and that of the missing-data indicators given the covariates. Parameters in the model are estimated by a pseudo-likelihood method. We first use the GEE with a logistic link to estimate the parameters associated with the marginal distribution of the missing-data indicator given the covariates, assuming that covariates are always observed. Then we estimate other parameters by inserting the estimates from the first step into the full likelihood function. A simulation study is conducted to assess the robustness of the assumed model under different missing-data processes. The proposed method is then applied to one example from a community cohort study to demonstrate its capability to reduce bias.

Aged↗
PubMed2001

Impact of abciximab versus eptifibatide on length of hospital stay for PCI patients.

The purpose of this study is to compare the profile of percutaneous coronary intervention (PCI) patients who receive abciximab versus eptifibatide, as well as to compare the effect of abciximab versus eptifibatide on hospital length of stay. Retrospective data were obtained from HCIA's Clinical Pathways Database on 5,446 coronary angioplasty patients who were administered either abciximab or eptifibatide. Estimation was conducted via a two-stage sample selection model. In the first stage, a probit regression was employed to determine which factors were associated with a higher probability of being administered abciximab versus eptifibatide. In the second stage, a negative binomial model was used to estimate the impact of a wide range of factors (selection of GPIIb/IIIa, patient demographics, insurance provider, health conditions, admission information, and hospital characteristics) on total hospital length of stay, as well as on postprocedural length of stay. After controlling for high-risk indications and other sources of selection bias, results indicate that receipt of abciximab was associated with a significantly shorter length of total hospital stay (0.83 fewer days; P < 0.001) than receipt of eptifibatide. Additionally, receipt of abciximab was found to be associated with a significantly shorter postprocedural hospital length of stay (0.48 fewer days; P = 0.002) compared to receipt of eptifibatide. Results of this study indicate that PCI patients who are administered abciximab versus eptifibatide have a significantly shorter length of hospital stay (both total and postprocedural). This finding is important since hospital length of stay reflects the occurrence of complications and has been found to be directly related to the resources consumed during in-patient management of patients. Cathet Cardiovasc Intervent 2001;53:296-303.

Abciximab↗
PubMed1997

Differential effects of B7-1 blockade in the rat experimental autoimmune encephalomyelitis model.

Blocking the CD28-B7 T cell costimulatory activation pathway protects animals from developing experimental autoimmune encephalomyelitis (EAE). In the mouse EAE model, selective blockade of B7-1 by specific mAbs has been shown to protect animals from EAE. In the Lewis rat model, we have shown that CD28-B7 blockade by systemic administration of CTLA4Ig prevents actively induced EAE. Since CTLA4Ig binds to both B7-1 and B7-2, we used a mutant form of CTLA4Ig (CTLA4IgY100F) that binds only B7-1, to study the role of B7-1 blockade in this model. Such a reagent avoids the potential of signaling by mAbs. Systemic administration of CTLA4IgY100F in several dosing regimens did not protect from EAE, and in some protocols worsened disease, while CTLA4Ig was always protective. In contrast, systemic injection of APCs preincubated ex vivo with the encephalitogenic peptide of myelin basic protein and either CTLA4Ig or CTLA4IgY100F protected recipients from disease. In vitro studies confirmed the in vivo observations and showed that primed lymph node cells from protected animals had decreased proliferative responses to myelin basic protein as compared with controls, while lymphocytes from animals treated with systemic CTLA4gY100F did not. More importantly, systemic administration of CTLA4IgY100F abrogated the protective effect of ex vivo treated APCs. These data suggest an important regulatory role for B7-1, perhaps through binding to CTLA4, in this model of EAE. Understanding the role and mechanisms of selective blockade of costimulatory molecules has implications for therapy of autoimmune disease.

Animals↗
PubMed1994

Rare alleles, MHC and captive breeding.

In recent years, more detailed genetic information has become available for individuals of endangered species in captive breeding programs. There have been suggestions that this information be used to identify rare alleles, particularly those at the MHC, that can be subsequently selected for captive breeding programs. First, we summarize the current information on the MHC relevant to conservation genetics, so that such a possible breeding program is seen in a proper perspective. For example, very few specific alleles at the MHC have been identified as selectively advantageous, even though there has been substantial effort to find such alleles in humans and a few other organisms. Further, many of the balancing selection models suggested for MHC variation are based on heterozygotes in general having a higher fitness than homozygotes and not on specific selectively advantageous alleles. Because there is no detailed data on MHC variability in captive populations, we used transferrin data in Przewalski's horses to evaluate a breeding program to select for rare alleles. In this species, one individual, 1060, has been identified to have the transferrin allele J. We determine the effect on founder contribution of multiply mating 1060 to increase the number of copies of this allele. Since there were 485 individuals in the population at this time, this extra mating had little detrimental effect on the distribution of founder contributions and the number of founder equivalents. We then selected 65, an ancestor of 1060, which had a high likelihood of being the individual that passed on the J allele in the lineage of 1060. We examined the effect of increasing the number of copies of alleles of 65 at a time when the population had only 22 other individuals. In this case, even though the founder contributions were changed more, there was also little effect on the founder contributions and the number of founder equivalents. Overall, it appears that selection that results in a limited change in the number of copies of rare alleles may not always have an overall detrimental effect. However, because other pedigrees may have very different properties, it is essential to perform a detailed pedigree analysis of any such selective breeding program to determine its effect before such a selection program is implemented.

Alleles↗
PubMed1997

Treatment-patient interactions for diagnostics of cross-over trials.

In cross-over trials, various types of responses may be recorded, not all of which can be appropriately modelled by a Normal distribution. Widening the class of models to the generalized linear model family has a number of advantages. An important one is that certain interactions, especially that between patients and treatments, can easily be fitted for frequency and count data. These can be used as diagnostics for the fit of the model used. One handicap has been the frequentist difficulty of comparing the fit of different non-nested models in this family. This can be overcome by the use of a model selection criterion such as the Akaike or Bayesian information criterion. This approach to modelling and diagnostics for cross-over trials is applied to two studies involving small counts of anginal attacks, previously analysed in the literature using classical Normal techniques.

Clinical Trials as Topic↗