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.

SourceResearch connectionFrom yearThrough year

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

At least 1,189 records · Page 66Linked to original sources

PubMed2002

Functional mixed effects models.

In this article, a new class of functional models in which smoothing splines are used to model fixed effects as well as random effects is introduced. The linear mixed effects models are extended to nonparametric mixed effects models by introducing functional random effects, which are modeled as realizations of zero-mean stochastic processes. The fixed functional effects and the random functional effects are modeled in the same functional space, which guarantee the population-average and subject-specific curves have the same smoothness property. These models inherit the flexibility of the linear mixed effects models in handling complex designs and correlation structures, can include continuous covariates as well as dummy factors in both the fixed or random design matrices, and include the nested curves models as special cases. Two estimation procedures are proposed. The first estimation procedure exploits the connection between linear mixed effects models and smoothing splines and can be fitted using existing software. The second procedure is a sequential estimation procedure using Kalman filtering. This algorithm avoids inversion of large dimensional matrices and therefore can be applied to large data sets. A generalized maximum likelihood (GML) ratio test is proposed for inference and model selection. An application to comparison of cortisol profiles is used as an illustration.

Bayes Theorem↗

PubMed2026

Diagnostic performance of machine learning models for malignant and non-malignant pleural effusion: Systematic review and meta-analysis.

BACKGROUND: Accurately distinguishing malignant pleural effusion (MPE) from non-malignant pleural effusion is clinically important, but the generalisability and methodological quality of machine-learning (ML) models remain uncertain. METHODS: We searched eight databases to 23 April 2026. Diagnostic performance was pooled using random-effects and Reitsma bivariate models, and study quality was assessed using PROBAST+AI. RESULTS: Forty-two studies were included; 17 contributed to the AUC meta-analysis and 14 to the bivariate analysis. The pooled AUC was 0.90 (95 % CI 0.85-0.94; 95 % prediction interval 0.62-0.98), with sensitivity of 0.80 (95 % CI 0.77-0.83) and specificity of 0.87 (95 % CI 0.79-0.92). Only nine studies reported external, temporal or independent validation. Externally validated studies had a lower pooled AUC than studies without external validation (0.83 vs 0.92), with lower specificity observed in the two externally validated studies contributing sensitivity and specificity data. All 42 development assessments had high overall quality concerns, and all 42 model evaluations were judged at high risk of bias. CONCLUSIONS: ML models showed good apparent accuracy for distinguishing MPE from non-MPE, but the evidence was limited by substantial heterogeneity, high risk of bias and scarce external validation. The pooled estimates reflect the average performance of different selected models rather than the expected accuracy of a single clinical test. ML models should be regarded as adjuncts to existing diagnostic pathways until they are confirmed by rigorous multicentre prospective external validation and clinical-impact studies.

Humans↗

PubMed1996

Identifying obstructive sleep apnea in patients presenting for laser-assisted uvulopalatoplasty.

The authors report on a series of 850 patients with snoring who were evaluated for laser-assisted uvulopalatoplasty (LAUP). Stepwise multivariate linear regression was employed to correlate patient symptoms and characteristics to the respiratory disturbance index (RDI). Body mass index, falling asleep while driving, snoring every night, and stopping breathing during sleep were found to correlate strongly with an increasing RDI (variance of 25%). Logistic multivariate linear regression analysis was used to predict the outcome of apnea (RDI>10). This model selected all of the above variables, as well as age, male sex, and the total number of symptoms, as being strong predictors of apnea. A receiver operating characteristic curve was used to describe the ability of this model to predict apnea. The authors conclude that otolaryngologists play an important role in the evaluation of OSAS, especially when a snoring patient may undergo LAUP, and they present an algorithm for the evaluation of such a patient. The authors believe that the clinical assessment, including a thorough history and a complete physical examination, remains extremely important in this evaluation. At present, the authors strongly recommend referral for a PSG if there is any suspicion of OSAS.

Adult↗

PubMed1978

Determinants of physician recognition and follow-up of abnormal laboratory values.

Potential determinants of physician recognition and follow-up of abnormal laboratory values were studied in the ambulatory primary care setting. Data support the hypothesis that a significant positive association exists between the clinical importance of a laboratory result and physician response. Clinical importance was indicated by degree of abnormality of the laboratory value, the type of test, and the indication for obtaining the test. Response was not significantly associated with type of laboratory report or resident's year of training, but a relationship was shown with resident's National Board scores. A model of 12 laboratory tests was found to be more appropriate for studying recognition and follow-up than one of 30 because of fewer repetitious tests and fewer results of doubtful clinical usefulness. With such a select model, recognition and follow-up of abnormals can be used as process measures of quality of medical care.

Adolescent↗

PubMed2001

Addition patterns, codes and contact graphs for fullerene derivatives.

The mathematical concept of the d-code and its associated contact graph give a model for sterically constrained addition patterns in fullerene derivatives C60Xm and C70Xm. In combination with simple electronic arguments, the stoichiometries, symmetries, and location of addends can be predicted, yielding a small number of candidates for further study. For example, sterically and optimal solutions C60Xm with pairwise separation of d bonds between addends are found at m(d) = 24(2), 12(3), 6(4,5), 2(6 to 9). The solution for C60X24 is unique, and the model selects 12 candidates for C60X12 from a starting set of 11661527060 possibilities.

Carbon↗

PubMed2003

Cell affinity separations using magnetically stabilized fluidized beds: erythrocyte subpopulation fractionation utilizing a lectin-magnetite support.

A magnetically stabilized fluidized bed is used to separate erythrocyte subpopulations. Binding specificity was obtained by immobilizing the lectin Helix pomatia Agglutinin (HpA) or Griffonia simplicifolia I (GSI) onto a magnetite-containing support. Separation of type A and type O erythrocytes with the lectin HpA was particularly effective, leading to a 94% purity of retained type A erythrocytes. A 3.1 +/- 0.6 log removal of type A erythrocytes was also accomplished leading to a 99.7% +/- 0.4% purity and 95% +/- 7% yield of type O erythrocytes in the collected effluent. Elution of the purified cells was accomplished using fluidization in the presence of a sugar competing for the lectin-erythrocyte binding site. A mathematical model based on the depth filtration model of Putnam and Burns (Chem Eng Sci 1997;52(1):93-105) was extended to include multicomponent cell adhesion. This filtration model is the first to take into account the finite binding capacity of the chromatographic support and is used to characterize the cell binding behavior and to determine optimal parameters and conditions that lead to high capacities and selectivities. Model parameter values and observations from in situ adsorption studies suggest that the non-spherical shape of the magnetite-based support allows for a more efficient utilization of the support surface area than the spherical shape. Using a 1.5-cm diameter laboratory column and realistic parameter values, the processing rates of the system are predicted to be at least an order of magnitude greater than the 10(8)/h cells that can typically be processed in packed bed cell affinity chromatography (CAC) systems.

ABO Blood-Group System↗

PubMed1995

Towards a general function describing T cell proliferation.

A new function is proposed for describing the rate of T cell proliferation in response to peptides on antigen-presenting cells. The model improves an earlier model of ours by allowing for a true maximum proliferation rate of the T cells. This is achieved by a simple change of variables that markedly relaxes the conditions for a conventional quasi-steady-state assumption. The new model has the same "ecological" properties as the previous one. Thus the natural competition in the model allows for regulation of T cell population size in the presence of continuous stimulation by antigen. An important feature is the competitive exclusion of T cell clones recognizing the same peptide with different affinities allowing for "affinity selection". Models for the population dynamics of experienced, naïve and activated T cells are also developed. These T cell subpopulations compete with one another for antigen. In models with lymphokine production a "proliferation threshold" is obtained that allows for tolerance.

Animals↗

PubMed2004

Risk factors associated with incident clinical vertebral and nonvertebral fractures in postmenopausal women: the Canadian Multicentre Osteoporosis Study (CaMos).

Utilizing data from the Canadian Multicentre Osteoporosis Study (CaMos), we examined the association between potential risk factors and incident vertebral and nonvertebral fractures. A total of 5,143 postmenopausal women were enrolled. Information collected during the study included data from the CaMos baseline and annually mailed fracture questionnaires, the Short Form 36 (SF-36), the Health Utilities Index, and physical measurements. Participants were followed for 3 years. Postmenopausal women were classified into four groups according to their incident fracture status since baseline: those without a new fracture; those with a new clinically recognized vertebral fracture; those with an incident nonvertebral fracture at the wrist, hip, humerus, pelvis, or ribs (main nonvertebral fracture group); and those with any new nonvertebral fracture (any-nonvertebral-fracture group). We performed multivariate Cox proportional hazard analysis using all possible risk factors to determine the association between risk factors and the time to the first minimal trauma fracture. Best predictive models were also determined using variables that were included in the full models. The Bayesian information criterion was used for model selection. For all analyses, relative risks and associated 95% confidence intervals were calculated. During the follow-up period, 34, 163, and 280 women developed a vertebral, a main nonvertebral, or any nonvertebral fracture, respectively. The best predictive models indicated that a five point lower quality of life as measured by the SF-36 physical component summary score was associated with relative risks of 1.21 (95% CI, 1.02 to 1.44), 1.17 (95% CI, 1.07 to 1.28), and 1.19 (95% CI, 1.11 to 1.27) for incident vertebral, main nonvertebral, and all nonvertebral fractures, respectively. In addition, for a one standard deviation (SD=0.12) lower femoral neck BMD, the relative risks for incident vertebral, main nonvertebral, and any nonvertebral fractures increased by 2.73 (95% CI, 1.74 to 4.28), 1.39 (95% CI, 1.06 to 1.82), and 1.34 (95% CI, 1.09 to 1.65), respectively. Furthermore, various anthropometric measures, disease conditions, and medications are associated with a new fracture. Identifying postmenopausal women at risk is important given that fracture prevention therapies are now available.

Adult↗

PubMed1999

Practical considerations in using calculated healthy-tissue complication probabilities for treatment-plan optimization.

PURPOSE: Healthy and neoplastic tissues are generally exposed nonuniformly to ionizing radiation. It is thus useful to develop algorithms that predict the probability of tumor control or normal tissue complication probability (NTCP) for any given spatial pattern of dose delivery. The questions addressed here concern: (a) the sensitivity of the NTCP predictions to the actual model used for extrapolation from uniform irradiation (where some clinical data exist) to nonuniform exposures, (b) its dependence on tissue type, and (c) consequences for treatment-plan optimization. METHODS AND MATERIALS: Two (of several possible) NTCP formulations are used here: the Lyman model and a binomial equation. The effective volume-reduction scheme of Kutcher and Burman is used to obtain the NTCP for an arbitrary distribution of dose. NTCP was calculated for seven organs by postulating a dose distribution of maximum nonuniformity. RESULTS: Both models fit available NTCP data well, but have very different extrapolations for exposures of small tissue volumes and very low values of NTCP (e.g., < 5%) where no data exist. Organs with pronounced volume effects (lung, kidneys) show substantial NTCP differences between the two models. Even in organs where the volume effect is small (e.g., spinal cord, brain), differences in NTCP due to the model selected may still have serious clinical consequences, as an actual example (for the spinal cord) indicates. CONCLUSIONS: NTCP calculations based on extrapolations to volume fractions and/or NTCP levels for which reliable data do not exist depend on the model used to fit the data and the degree of dose nonuniformity. If NTCP is to be used in treatment-plan optimization, the prudent approach is to design plans that reproduce the conditions under which available dose-volume data were taken (e. g., uniform dose distributions).

Algorithms↗

PubMed2006

Small mammals and forest fuel reduction: national-scale responses to fire and fire surrogates.

Forest fuel reduction treatments are increasingly used by managers to reduce the risk of high-severity wildfire and to manage changes in the ecological function of forests. However, comparative ecological effects of the various types of treatments are poorly understood. We examined short-term patterns in small-mammal responses to mechanical thinning, prescribed-fire, and mechanical thinning/prescribed-fire combination treatments at eight different study areas across the United States as a part of the National Fire and Fire Surrogate (FFS) Project. Research questions included: (1) do treatments differ in their effect on small mammal densities and biomass? and (2) are effects of treatments consistent across study areas? We modeled taxa-specific densities and total small-mammal biomass as functions of treatment types and study area effects and ranked models based on an information-theoretic model selection criterion. Small-mammal taxa examined, including deer mice (Peromyscus maniculatus), yellow-pine chipmunks (Tamias amoenus), and golden-mantled ground squirrels (Spermophilus lateralis), as well as all Peromyscus and Tamias species, had top-ranked models with responses varying both by treatment type and study area. In each of these cases, the top-ranked model carried between 69% and 99% of the total weight in the model set, indicating strong support for the top-ranked models. However, the top-ranked model of total small-mammal biomass was a model with biomass varying only with treatment (i.e., treated vs. untreated), not by treatment type or study area; again, this model had strong support, with 75% of the total model weight. Individual species and taxa appear to have variable responses to fuel reduction treatment types in different areas; however, total small-mammal biomass appears generally to increase after any type of fuel reduction. These results suggest that there is substantial variability in taxa-specific responses to treatments and indicate that adaptive management policies may be necessary when applying fuel reduction treatments in areas where management of small-mammal populations is of interest. Adaptive management can be used by managers who are conducting fuel reduction treatments to reduce uncertainty as to which treatments are locally optimal for meeting objectives for the management of small-mammal populations.

Animals↗

PubMed2003

Maximum likelihood methods for nonignorable missing responses and covariates in random effects models.

This article analyzes quality of life (QOL) data from an Eastern Cooperative Oncology Group (ECOG) melanoma trial that compared treatment with ganglioside vaccination to treatment with high-dose interferon. The analysis of this data set is challenging due to several difficulties, namely, nonignorable missing longitudinal responses and baseline covariates. Hence, we propose a selection model for estimating parameters in the normal random effects model with nonignorable missing responses and covariates. Parameters are estimated via maximum likelihood using the Gibbs sampler and a Monte Carlo expectation maximization (EM) algorithm. Standard errors are calculated using the bootstrap. The method allows for nonmonotone patterns of missing data in both the response variable and the covariates. We model the missing data mechanism and the missing covariate distribution via a sequence of one-dimensional conditional distributions, allowing the missing covariates to be either categorical or continuous, as well as time-varying. We apply the proposed approach to the ECOG quality-of-life data and conduct a small simulation study evaluating the performance of the maximum likelihood estimates. Our results indicate that a patient treated with the vaccine has a higher QOL score on average at a given time point than a patient treated with high-dose interferon.

Analysis of Variance↗

PubMed2005

Comparison of three comorbidity measures for predicting health service use in patients with osteoarthritis.

OBJECTIVE: To compare the ability of 3 database-derived comorbidity scores, the Charlson Score, Elixhauser method, and RxRisk-V, in predicting health service use among individuals with osteoarthritis (OA). METHODS: The study population comprised 306 patients who were under care for OA in the Veterans Affairs (VA) health care system. Comorbidity scores were calculated using 1 year of data from VA inpatient and outpatient databases (Charlson Score, Elixhauser method), as well as pharmacy data (RxRisk-V). Model selection was used to identify the best comorbidity index for predicting 3 health service use variables: number of physician visits, number of prescriptions used, and hospitalization probability. Specifically, Akaike's Information Criterion (AIC) was used to determine the best model for each health service outcome variable. Model fit was also evaluated. RESULTS: All 3 comorbidity indices were significant predictors of each health service outcome (P < 0.01). However, based on AIC values, models using the RxRisk-V and Elixhauser indices as predictor variables were better than models using the Charlson Score. The model using the RxRisk-V index as a predictor was the best for the outcome of prescription medication use, and the model with the Elixhauser index was the best for the outcome of physician visits. CONCLUSION: The Rx-Risk-V and Elixhauser are suitable comorbidity measures for examining health services use among patients with OA. Both indices are derived from administrative databases and can efficiently capture comorbidity among large patient populations.

Comorbidity↗

PubMed2001

Sequence analysis using logic regression.

Logic Regression is a new adaptive regression methodology that attempts to construct predictors as Boolean combinations of (binary) covariates. In this paper we use this algorithm to deal with single-nucleotide polymorphism (SNP) sequence data. The predictors that are found are interpretable as risk factors of the disease. Significance of these risk factors is assessed using techniques like cross-validation, permutation tests, and independent test sets. These model selection techniques remain valid when data is dependent, as is the case for the family data used here. In our analysis of the Genetic Analysis Workshop 12 data we identify the exact locations of mutations on gene 1 and gene 6 and a number of mutations on gene 2 that are associated with the affected status, without selecting any false positives.

Algorithms↗

PubMed1998

Analysis of genetic structure and dispersal patterns in a population of sea beet.

A model of the migration pattern in a metapopulation of sea beet (Beta vulgaris L. ssp. maritima), based on the continuous distributions of seed and pollen movements, is fitted to gene frequency data at 12 isozyme and RFLP loci by maximum likelihood by using an approximation of the simultaneous equilibrium distribution of the gene frequencies generated by the underlying multivariate stochastic process of genetic drift in the population. Several alternative restrictions of the general model are fitted to the data, including the island model, a model of complete isolation, and a model in which the seed and pollen dispersal variances are equal. Several likelihood ratio tests between these alternatives are performed, and median bias in the estimated parameters is corrected by using parametric bootstrapping. To assess the fit of the selected model, the predicted covariances are compared with covariances computed from the data directly. The dependency of estimated parameters on the ratio between effective and absolute subpopulation sizes, which is treated as a known parameter in the analysis, is also examined. Finally, we note that the data also appear to contain some information about this ratio.

Alleles↗

PubMed2004

Genomic approaches to understanding mammary tumor progression in transgenic mice and responses to therapy.

Scores of genetically engineered mice have been generated in the quest to understand mechanisms of breast cancer development and progression. More recently, there has been a growing trend for using such models for testing various therapeutic strategies and agents. The application of these mouse models for these purposes requires that they be characterized in ways that demonstrate they possess important similarities to human breast cancer. In particular, detailed comparisons of the features of the models to human breast cancer must include attention to the histological phenotypes, chromosomal and molecular alterations, and the predictive value of the models for preclinical testing. Whereas these models have become important tools for the study of breast cancer, the great majority of existing mouse mammary cancer models develop tumors that are estrogen receptor negative, with relatively few models demonstrating metastatic spread to the lungs, and none developing metastases to bone. This review focuses on recent studies using genomic approaches to further understand the oncogenic processes occurring in mouse models of mammary cancer and to compare these changes with those identified in human breast cancer. Gene expression profiling is being applied to help define pharmacological responses that occur in vivo. Detailed genomic analyses will provide important information for selecting models for specific experimental purposes, contribute to the understanding of oncogene-specific expression signatures and potential therapeutic targets, and further define mechanisms of chemoprevention and chemotherapy.

Animals↗

PubMed2006

Beyond statistical inference: a decision theory for science.

Traditional null hypothesis significance testing does not yield the probability of the null or its alternative and, therefore, cannot logically ground scientific decisions. The decision theory proposed here calculates the expected utility of an effect on the basis of (1) the probability of replicating it and (2) a utility function on its size. It takes significance tests--which place all value on the replicability of an effect and none on its magnitude--as a special case, one in which the cost of a false positive is revealed to be an order of magnitude greater than the value of a true positive. More realistic utility functions credit both replicability and effect size, integrating them for a single index of merit. The analysis incorporates opportunity cost and is consistent with alternate measures of effect size, such as r2 and information transmission, and with Bayesian model selection criteria. An alternate formulation is functionally equivalent to the formal theory, transparent, and easy to compute.

Data Interpretation, Statistical↗

PubMed1999

A canine model of heart failure by intracoronary adriamycin injection: hemodynamic and energetic results.

BACKGROUND: To eliminate the noncardiac toxicities associated with Adriamycin (Pharmacia Upjohn, Kalamazoo, MI) and to avoid the need for surgical implantation of a coronary artery catheter, we proposed serial intracoronary infusions of Adriamycin (Pharmacia Upjohn) with a Judkins catheter to develop a nonsurgical but selective model of dilated cardiomyopathy in dogs. METHODS AND RESULTS: Twenty-two dogs were entered onto the study. Each dog received four weekly injections of Adriamycin (Pharmacia Upjohn; 15 mg/wk) through a Judkins catheter and verapamil orally (5 mg/kg twice daily) for 4 weeks. Dogs were evaluated with echocardiography, right-sided cardiac catheterization, and endomyocardial biopsy at times 0 and 10 weeks. The left ventricular ejection fraction decreased from 0.65 +/- 0.04 to 0.48 +/- 0.11 (P = .0006). The left ventricular end-diastolic and end-systolic diameter indices increased from 42.00 +/- 2.17 to 46.92 +/- 5.81 mm (P = .0029) and from 27.65 +/- 2.59 to 35.00 +/- 6.84 mm, respectively (P = .0210). The mechanical cardiac efficiency decreased from 28.55% +/- 13.30% at baseline and 15.64% +/- 9.09% at 10 weeks (P = .001). Cardiac myocyte vacuolation was present on histological examination at 10 weeks. The mortality rate was 41%. Minimal systemic Adriamycin (Pharmacia Upjohn) toxicity was noticed. CONCLUSION: Intracoronary injection of Adriamycin (Pharmacia Upjohn) in dogs induced a chronic dilated cardiomyopathy.

Animals↗

PubMed1995

Migrant selectivity and returns to skills: the case of Taiwanese immigrants in the United States.

"This article examines the selection of international migration from Taiwan to the United States in terms of the skills and abilities of emigrants. Micro data from Taiwan and the U.S. are used to test Borjas' self-selection model (1987).... Based on the assumption that emigrants choose to move to the destination that can maximize their expected lifetime earnings and provide higher returns to their human capital, Heckman (1979) developed a two-stage method that allows one to correct the problem of selectivity bias and to estimate directions of the selection on the basis of observed characteristics (the levels of skills or education) and unobserved characteristics (innate ability or motivation) for both emigrants and non-emigrants." (SUMMARY IN FRE AND SPA)

Americas↗