Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “statistical inference”

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 1,585 records · Page 88Linked to original sources
PubMed2000

Exponential family models and statistical genetics.

This article describes the evolution of applied exponential family models, starting at 1972, the year of publication of the seminal papers on generalized linear models and on Cox regression, and leading to multivariate (i) marginal models and inference based on estimating equations and (ii) random effects models and Bayesian simulation-based posterior inference. By referring to recent work in genetic epidemiology, on semiparametric methods for linkage analysis and on transmission/disequilibrium tests for haplotype transmission this paper illustrates the potential for the recent advances in applied probability and statistics to contribute to new and unified tools for statistical genetics. Finally, it is emphasized that there is a need for well-defined postgraduate education paths in medical statistics in the year 2000 and thereafter.

Biometry↗
PubMed2026

Genetic Correlation Between Brain Imaging Phenotypes and Externalizing Behavior: A Large-Scale LDSC Analysis of UK Biobank IDPs.

Externalizing has been associated with differences in brain structure and function; however, it remains unclear whether these associations reflect shared common-variant genetic influences. Cross-trait linkage disequilibrium score regression was used to estimate genome-wide genetic correlations between externalizing genome-wide association study (GWAS) results and 3,935 brain imaging-derived phenotypes from the UK Biobank BIG40 resource. The imaging phenotypes covered structural magnetic resonance imaging (MRI), diffusion MRI, susceptibility-weighted imaging, resting-state functional MRI, and task-based functional MRI. Results were included in the primary analysis when the imaging phenotype had positive single-nucleotide polymorphism (SNP) heritability, a heritability Z statistic of at least 1.96, a mean GWAS chi-square statistic of at least 1.02, at least 200,000 regression SNPs, and a complete LDSC result without a fatal error. Technical imaging quality-control phenotypes were excluded from biological inference. Individual results were corrected using the Benjamini-Hochberg false discovery rate procedure. Aggregated Cauchy association tests (ACATs) were used to evaluate evidence across all imaging phenotypes and within predefined imaging categories. Statistical power, simultaneous confidence bounds, and alternative quality-control definitions were examined in sensitivity analyses. Of the 3,935 imaging phenotypes, 3,716 produced estimable genetic correlations, 2,980 met the primary LDSC quality-control criteria, and 2,967 were classified as biological imaging phenotypes. No individual phenotype survived false discovery rate correction. The smallest unadjusted P value was 0.0005, and the minimum adjusted q value was 0.486. The distribution of genetic correlations was centered near zero, with a median genetic correlation of 0.0014 and a median absolute genetic correlation of 0.0338. ACAT provided no evidence of an aggregate association across all biological imaging phenotypes (P = 0.302), and no predefined imaging category survived multiple-testing correction. The median minimum detectable genetic correlation at 80% power was 0.216. Bonferroni-adjusted simultaneous confidence intervals were fully contained within the interval [-0.30, 0.30] for 80.0% of phenotypes in the primary analysis and 88.0% under the stringent heritability quality-control definition. Broad and stringent sensitivity analyses produced the same overall conclusions. In this study, no statistically robust evidence of genome-wide genetic correlations between externalizing and individual UK Biobank brain imaging phenotypes was found. Nevertheless, small, localized, mixed-direction, or developmentally specific genetic effects remain possible.

Journal Article↗
PubMed1998

On the discrepancy between epidemiologic studies in individuals of lung cancer and residential radon and Cohen's ecologic regression.

There is still substantial confusion in the radiation effects community about the inherent limitations of ecologic analysis. As a result, inordinate attention has been given to the discrepant results of Cohen, in which a negative estimate is observed for the regression of county mortality rates for lung cancer on estimated county radon levels. This paper demonstrates that Cohen's ecologic analysis cannot produce valid inference on the exposure-response relationship for individuals unless lung cancer risk factors (smoking, age, occupation, etc.) for individuals are statistically uncorrelated with indoor radon level within counties or unless risk effects for radon and other factors are additive. Both of these assumptions are contradicted in the literature. Thus, contrary to common assumption, when a linear no-threshold model is the true model for radon risk for individuals, higher average radon concentration for a county does not necessarily imply a higher lung cancer rate for the county. In addition, valid inference from county-level ecologic analysis and the elimination of the ecologic bias cannot be achieved with the addition of county-wide summary variables (including "stratification" variables) to the regression equation. Using hypothetical data for smoking and radon and assuming a true positive association for radon and lung cancer for individuals, the analysis demonstrates that a negative county-level ecologic regression can be induced when correlation coefficients for smoking and radon within county are in the range -0.05 to 0.05. Since adverse effects for radon at low exposures are supported by analysis of miner data (all data and data restricted only to low cumulative exposures), a meta-analysis of indoor radon studies, and molecular and cellular studies, and since ecologic regressions are burdened by severe limitations, the negative results from Cohen's analysis are most likely due to bias and should be rejected.

Air Pollutants, Radioactive↗
PubMed2002

Empirical bayes methods and false discovery rates for microarrays.

In a classic two-sample problem, one might use Wilcoxon's statistic to test for a difference between treatment and control subjects. The analogous microarray experiment yields thousands of Wilcoxon statistics, one for each gene on the array, and confronts the statistician with a difficult simultaneous inference situation. We will discuss two inferential approaches to this problem: an empirical Bayes method that requires very little a priori Bayesian modeling, and the frequentist method of "false discovery rates" proposed by Benjamini and Hochberg in 1995. It turns out that the two methods are closely related and can be used together to produce sensible simultaneous inferences.

Bayes Theorem↗
PubMed2006

A scan statistic for identifying chromosomal patterns of SNP association.

We have developed a single nucleotide polymorphism (SNP) association scan statistic that takes into account the complex distribution of the human genome variation in the identification of chromosomal regions with significant SNP associations. This scan statistic has wide applicability for genetic analysis, whether to identify important chromosomal regions associated with common diseases based on whole-genome SNP association studies or to identify disease susceptibility genes based on dense SNP positional candidate studies. To illustrate this method, we analyzed patterns of SNP associations on chromosome 19 in a large cohort study. Among 2,944 SNPs, we found seven regions that contained clusters of significantly associated SNPs. The average width of these regions was 35 kb with a range of 10-72 kb. We compared the scan statistic results to Fisher's product method using a sliding window approach, and detected 22 regions with significant clusters of SNP associations. The average width of these regions was 131 kb with a range of 10.1-615 kb. Given that the distances between SNPs are not taken into consideration in the sliding window approach, it is likely that a large fraction of these regions represents false positives. However, all seven regions detected by the scan statistic were also detected by the sliding window approach. The linkage disequilibrium (LD) patterns within the seven regions were highly variable indicating that the clusters of SNP associations were not due to LD alone. The scan statistic developed here can be used to make gene-based or region-based SNP inferences about disease association.

Chromosome Mapping↗
PubMed1988

Individual differences and variability in the timing of motor activity during walking in insects.

The uniformity of the neural physiology of an animal population is a fundamental, rarely tested assumption in most neurophysiological work. In this study, the variability of the timing between the movements of pairs of legs during free walking in cockroaches was assessed. Phases (a measure of timing) of motor bursts in muscles of legs in the American cockroach, Periplaneta americana, were calculated for insects walking straight over a flat, level surface. Student's t, Wallraff, Mann Whitney and Watson U2 two-sample tests were used to compare the phases of motor bursts of the same pairs of legs in different insects. The comparisons showed that in spite of the homogeneity both of the animal population and of the conditions under which the insects walked, most of the inter-leg phases of the animals that were compared were significantly different statistically. Further testing of greater numbers of insects using analysis of variance to test for population uniformity confirmed that the insects we tested were not members of a single statistical population with respect to the timing of motor bursts of the legs during walking. We infer that this unexpectedly large variability in a population thought to be relatively homogeneous reflects subtle but biologically significant differences between animals. The possible sources of these differences and their consequences for the study of behavior and its physiological basis are discussed.

Animals↗
PubMed1993

Bayesian inference for model-based segmentation of computed radiographs of the hand.

We present a method for medical image understanding by computer that uses model-based, hierarchical Bayesian inference to accurately segment imaged anatomy. A first application is a prototype system that automatically segments and measures symptoms of arthridities in hand radiographs. This is potentially useful in radiological diagnosis and tracking of arthridities. Key steps of the model-based, Bayesian inference approach are: (1) prediction of imagery features from 3D models of anatomy, parameterized by population statistics, (2) local image feature extraction in predicted sub-regions, and (3) the use of a probabilistic calculus to accrue results of image processing and image feature matching procedures in support or denial of hypotheses about the imaged anatomy. The prototype system for hand radiograph analysis accurately segments normal and somewhat degenerated hand anatomy. Results are shown of the ability of the automated system to 'fail soft', recognizing when segmentation is inadequate for accurate measurement. This self evaluation capability improves reliability of measurements for potential clinical use.

Arthritis↗
PubMed2004

Phylogeny and evolution of basils and allies (Ocimeae, Labiatae) based on three plastid DNA regions.

A phylogeny of basils and allies (Lamiaceae, tribe Ocimeae) based on sequences of the trnL intron, trnL-trnF intergene spacer and rps 16 intron of the plastid genome is presented. Several methods were used to reconstruct phylogenies and to assess statistical support for clades: maximum parsimony with equally and successively weighted characters, bootstrap resampling, and Bayesian inference. The phylogeny is used to investigate the distribution of morphological, pericarp anatomy, chemical, and pollen characters as well as the geographical distribution of the clades. Tribe Ocimeae is monophyletic and easily diagnosable with morphological synapomorphies. There are monophyletic clades within Ocimeae that broadly correspond to currently recognised subtribes: Lavandulinae, Hyptidinae, Ociminae, and Plectranthinae. Only Lavandulinae has clear non-molecular synapomorphies. Several currently recognised genera are not monophyletic. Floral morphology consistent with sternotribic pollination is most common in Ocimeae, but there are independent departures from this model. Buzz pollination is likely in some species, the only postulated occurrence of this within Lamiaceae. Quinone diterpenoids and flavones in the leaf exudates differ in their distributions across the phylogeny and this could contribute to differences in the recorded medicinal as well as pesticidal uses of the species in the different clades. Mapping geographic distribution on to an ultrametric phylogenetic tree produced using non-parametric rate smoothing supports an Asiatic origin for Ocimeae. There are several secondary occurrences in Asia arising from the African Ociminae and Plectranthinae clades. Colonisation of Madagascar occurred at least five times, and New World colonisation occurred at least three times.

Bayes Theorem↗
PubMed2006

Gastrointestinal helminths in indigenous and exotic chickens in Vietnam: association of the intensity of infection with the Major Histocompatibility Complex.

This study compared the prevalence and intensity of infections of helminths in 2 chicken breeds in Vietnam, the indigenous Ri and the exotic Luong Phuong. Also, possible correlations with the Major Histocompatibility Complex (MHC) were tested. The most prevalent helminths were Ascaridia galli, Heterakis beramporia, Tetrameres mothedai, Capillaria obsignata, Raillietina echinobothrida and Raillietina tetragona. Differences in prevalence and intensity of infection were found between the 2 breeds. Comparing the 2 groups of adult birds, Ri chickens were observed to have higher prevalence and infection intensities of several species of helminths, as well as a higher mean number of helminth species. In contrast, A. galli and C. obsignata were shown to be more prevalent in Luong Phuong chickens. Furthermore, an age-dependent difference was indicated in the group of Ri chickens in which the prevalence and the intensity of infection was higher for the adult than the young chickens for most helminths. The most notable exception was the significantly lower prevalence and intensities of A. galli in the group of adult chickens. In contrast, the prevalence and intensity were very similar in both age groups of Luong Phuong chickens. Using a genetic marker located in the MHC, a statistically significant correlation between several MHC haplotypes and the infection intensity of different helminth species was inferred. This is the first report of an association of MHC haplotype with the intensity of parasite infections in chickens.

Animals↗
PubMed2002

Insufficiently dehydrated hydrogen bonds as determinants of protein interactions.

The prediction of binding sites and the understanding of interfaces associated with protein complexation remains an open problem in molecular biophysics. This work shows that a crucial factor in predicting and rationalizing protein-protein interfaces can be inferred by assessing the extent of intramolecular desolvation of backbone hydrogen bonds in monomeric structures. Our statistical analysis of native structures shows that, in the majority of soluble proteins, most backbone hydrogen bonds are thoroughly wrapped intramolecularly by nonpolar groups except for a few ones. These latter underwrapped hydrogen bonds may be dramatically stabilized by removal of water. This fact implies that packing defects are "sticky" in a way that decisively contributes to determining the binding sites for proteins, as an examination of numerous complexes demonstrates.

Desiccation↗
PubMed1984

Forgotten moments: a note on skewness and kurtosis as influential factors in inferences extrapolated from response distributions.

It is proposed that reliance on only the mean and standard deviation of a distribution to describe response frequency may lead to erroneous inferences concerning such distributions when skewness and kurtosis are present. After defining the first four moments of a distribution, it is demonstrated analytically that skewness and kurtosis may vary to systematically influence the mean and standard deviation of a set of related distributions. The significance of these relationships for the interpretation of differing response distributions is advanced through examples gleaned from the movement control literature. in addition, it is suggested that the use of bandwidths to select scores from a distribution for subsequent data analysis can further compound the problems of both descriptive and explanatory inference, particularly when skewness and kurtosis are features of such distribution(s). Whether or not inferential statistics are invoked, a veridical perspective of distributions is essential to meaningful data analysis.

Journal Article↗
PubMed2001

Methodology for treatment evaluation in patients with cancer metastatic to bone.

BACKGROUND: Patients with cancer metastatic to bone experience several adverse and clinically important skeletal-related events, including pathologic fractures, vertebral compressions with fracture, the need for surgery to treat or prevent fractures, and the need for radiation therapy for the treatment of bone pain. We present appropriate methods for describing and modeling the clinical course of skeletal-related events and comparing treatments for such events. METHODS: On the basis of data from a recently completed randomized, placebo-controlled trial involving 380 breast cancer patients with bone metastases, we tested the validity of the "events-per-person-years" method, one of the most commonly used techniques, for the analysis of skeletal-related events. We then used more robust methods of analysis that are based on fewer assumptions, including a random-effects Poisson model, and contrasted the inferences about skeletal-related event rates and treatment effects for the different analytic methods. All statistical tests were two-sided. RESULTS: The events-per-person-years analysis underestimated substantially the variation in the data and is not appropriate to summarize the incidence rate of skeletal-related events. A random-effects Poisson model did provide a valid basis for analyzing such data. CONCLUSIONS: The underestimation of variability in data associated with the use of the events-per-person-years analysis leads to unduly narrow confidence intervals for complication rates and inflated false-positive error rates in treatment comparisons. A random-effects Poisson model provides a valid, robust basis for describing the clinical course of bone complications and evaluating treatment effects.

Bone Neoplasms↗
PubMed2005

Evolutionary dynamics of human retroviruses investigated through full-genome scanning.

To test hypotheses on the differences in retroviral genetic diversity, we compared the evolutionary dynamics of the human immunodeficiency virus type 1 (HIV-1) group M and the primate T-cell lymphotropic virus (PTLV) using a full-genome analysis. Evolutionary rates and nonsynonymous/synonymous substitution rate ratios were estimated across the genome using a maximum likelihood sliding window approach, and molecular clock properties were investigated. We confirm a remarkable difference in genetic stability and selective pressure at the interhost level. While there is evidence for adaptive evolution in HIV-1, the evolution of PTLV is almost exclusively characterized by negative selection or nearly neutral processes. For both retroviruses, evolutionary rate estimates across the genome reflect the differential selective constraints. However, based on the relationship between evolutionary rate and selective pressure and based on the comparison of synonymous substitution rates, the differences in rate between HIV-1 and PTLV cannot be explained by selective forces only. Several evolutionary and statistical assumptions, examined using a Bayesian coalescent method, were shown to have little influence on our inference.

Bayes Theorem↗
PubMed1998

Longitudinal changes in lung function associated with aspects of swine-confinement exposure.

Several aspects of swine-confinement farming appear to be leading to adverse respiratory effects. This study was set up in a longitudinal design to study the association between certain characteristics of farms or the way they are run and a decline in lung function. A cohort of 171 pig farmers was observed for 3 years. Lung function was measured. Exposure to farm characteristics was determined at the start of the observation period, using data from standardized farm surveys and from diaries kept by the participants. Mean decline in lung function was 73 mL/year for forced expiratory volume in 1 second (FEV1) and 55 mL/year for forced vital capacity (FVC). A longitudinal decline in FEV1 was significantly associated with the use of quaternary ammonium compounds as disinfectants (an additional 43 mL/year) and also with the use of an automated dry feeding system (an additional 28 mL/year). The association with the use of wood shavings as bedding material was not statistically significant. The impact of these characteristics in a longitudinal study provides stronger evidence for causal inference than that shown in previous cross-sectional designs. This may be useful in promoting preventive measures.

Adult↗
PubMed1999

Valid inference in random effects meta-analysis.

The standard approach to inference for random effects meta-analysis relies on approximating the null distribution of a test statistic by a standard normal distribution. This approximation is asymptotic on k, the number of studies, and can be substantially in error in medical meta-analyses, which often have only a few studies. This paper proposes permutation and ad hoc methods for testing with the random effects model. Under the group permutation method, we randomly switch the treatment and control group labels in each trial. This idea is similar to using a permutation distribution for a community intervention trial where communities are randomized in pairs. The permutation method theoretically controls the type I error rate for typical meta-analyses scenarios. We also suggest two ad hoc procedures. Our first suggestion is to use a t-reference distribution with k-1 degrees of freedom rather than a standard normal distribution for the usual random effects test statistic. We also investigate the use of a simple t-statistic on the reported treatment effects.

Anticholesteremic Agents↗