Hierarchical Model Selection Research Articles

Detecting the Undetected: Overcoming Biases in Gravitational-wave Population Studies

In the flourishing field of gravitational-wave astronomy, accurately inferring binary black hole merger formation channels is paramount. The Bayesian hierarchical model selection analysis offers a promising methodology (see, e.g., “One Channel to Rule Them All”). However, recently, Cheng et al. highlighted a critical caveat: observed channels absent in known models can bias branching fraction estimates. In this research note, we introduce a test to detect missing channels in such analyses. Our findings show a commendable success rate in identifying these elusive channels. Yet, in scenarios where missing channels closely overlap with recognized ones, discerning the difference remains challenging.

Hierarchical Model Selection for Graph Neural Networks

Node classification on graph data is a major problem in machine learning, and various graph neural networks (GNNs) have been proposed. Variants of GNNs such as H2GCN and CPF outperform graph convolutional networks (GCNs) by improving on the weaknesses of the traditional GNN. However, there are some graph data which these GNN variants fail to perform well than other GNNs in the node classification task. This is because H2GCN has similar feature values on graph data with high average degree, and CPF gives rise to a problem about label-propagation suitability. Accordingly, we propose a hierarchical model selection framework (HMSF) that selects an appropriate GNN model to predict the class of nodes for each graph data. HMSF uses average degree and edge homophily ratio as indicators to decide the useful model based on our analyses. In the experiment, we show that the model selected by our HMSF achieves high performance on node classification for various types of graph data.

Symbolic sequence representation with Markovian state optimization

Sequence representation, which is aimed at embedding sequentially symbolic data in a real space, is a foundational task in sequence pattern recognition. It is a difficult problem due to the challenges entailed in learning the intrinsic structural features within sequences in small sample size cases, in an unsupervised way. In this paper, we propose to represent each symbolic sequence by its transition probability distribution over discriminating topics, formalized by a set of optimized Hidden Markov Model (HMM) states shared by all sequences. An efficient method, called Markovian state clustering with hierarchical model selection, is proposed to optimize the Markovian states and to adaptively determine the number of topics. The proposed method is experimentally evaluated on human activity recognition and protein recognition, and results obtained demonstrate its effectiveness and efficiency.

Increase in rear-end collision risk by acute stress-induced fatigue in on-road truck driving.

Increasing road crashes related to occupational drivers’ deteriorating health has become a social problem. To prevent road crashes, warnings and predictions of increased crash risk based on drivers’ conditions are important. However, in on-road driving, the relationship between drivers’ physiological condition and crash risk remains unclear due to difficulties in the simultaneous measurement of both. This study aimed to elucidate the relationship between drivers’ physiological condition assessed by autonomic nerve function (ANF) and an indicator of rear-end collision risk in on-road driving. Data from 20 male truck drivers (mean ± SD, 49.0±8.2 years; range, 35–63 years) were analyzed. Over a period of approximately three months, drivers’ working behavior data, such as automotive sensor data, and their ANF data were collected during their working shift. Using the gradient boosting decision tree method, a rear-end collision risk index was developed based on the working behavior data, which enabled continuous risk quantification. Using the developed risk index and drivers’ ANF data, effects of their physiological condition on risk were analyzed employing a logistic quantile regression method, which provides wider information on the effects of the explanatory variables, after hierarchical model selection. Our results revealed that in on-road driving, activation of sympathetic nerve activity and inhibition of parasympathetic nerve activity increased each quantile of the rear-end collision risk index. The findings suggest that acute stress-induced drivers’ fatigue increases rear-end collision risk. Hence, in on-road driving, drivers’ physiological condition monitoring and ANF-based stress warning and relief system can contribute to promoting the prevention of rear-end truck collisions.

Factors explaining the composition and diversity of vascular plant understories along a transcontinental gradient in the Canadian boreal forest

AbstractAimImpacts of longitudinal gradients on species distributions have received less attention than latitudinal gradients. In Canada, precipitation varies longitudinally, with direct effects on plant growth and fire regimes. Despite the geographical extent of Canadian boreal forests, vascular plant diversity is relatively limited, with just under 300 species. Understorey communities comprise most of this diversity and play key roles in forests, affecting succession, nutrient cycling and wildlife habitat. Our objective was to evaluate the relative impacts of local and regional environmental conditions on vascular plant community composition and diversity in the boreal forest.LocationSampling took place in 33 trembling aspen (Populus tremuloides) stands along a transcontinental gradient in precipitation and fire regimes in boreal Canada.MethodsWe measured community composition and vegetation diversity indices of the vascular understorey in 400‐m2 plots and measured environmental variables at the local (drainage, nutrient and light availability, site heterogeneity) and regional (annual temperature and precipitation, fire weather index, landscape heterogeneity) scales. Multivariate analyses and hierarchical model selection were used to analyze patterns.ResultsSpecies composition varied longitudinally, with western and eastern species pools. Western indicator species tended to be fire‐adapted pioneer species, while indicator species at eastern sites were fire avoiders and late‐successional species. Species richness and intra‐site beta diversity seemed primarily driven by local variables and did not vary across the country. Species evenness, however, was slightly higher in the western region and decreased under higher precipitation, colder temperatures, and higher landscape‐level heterogeneity.ConclusionOur results suggested that even for similar canopy composition environmental variables play a key role in the establishment of plant communities and structure local plant assemblages by selecting or eliminating species from the regional pool, which was also controlled by fire regimes.

Patterns of Monarch Site Occupancy and Dynamics in Iowa

The monarch butterfly (Danaus plexippus) is the focus of large-scale habitat restoration efforts because of recent population declines. From 2006-2017 we monitored monarchs at >400 sites throughout Iowa to link site occupancy and colonization/extinction dynamics to the presence of milkweed, site-specific habitat metrics, and landscape context at differing spatial scales. We used a robust design occupancy model in Program MARK and a hierarchical model selection approach to estimate site occupancy, extinction and colonization probabilities, and detection probability. Occupancy models revealed that monarchs responded differently to landscape features, environmental conditions, and local habitat conditions for site occupancy, extinction, and colonization probabilities. For site occupancy, the mean patch size of grassland at the 1-km spatial scale had a positive effect (β_GrassPS1K= 0.94, SE = 0.54) while the percent of the landscape in woodland at the 200-m spatial scale had a negative effect (β_WoodPL200= -1.68, SE = 0.34). For extinction, there were additive effects of the percent of the landscape in woodland at the 100-m spatial scale (β_WoodPLAND100= 2.70, SE = 0.63), the interspersion of grassland at the 1-km spatial scale (β_GrassIJI1K= -2.30, SE = 0.63), and litter depth (β_Litter= 0.46, SE = 0.13). Finally, there were negative effects of the percent of the landscape in woodland at the 200-m spatial scale (β_WoodPLAND200= -4.67, SE = 1.37) and the interspersion of grassland at the 100-m spatial scale (β_GrassPS1K= -2.02, SE = 0.70) on colonization probability. Detection probability was affected by the additive effects of canopy cover and monarch density; no other detection model was competitive. In the top model there was a positive effect of monarch density (β_Density = 0.28, SE = 0.05) and a negative effect of canopy cover (β_Canopy=-0.18, SE = 0.03) on detection probability. In Iowa, monarchs are widespread on conservation lands where they avoid sites with lots of canopy cover. Colonization and extinction processes are influenced by an interplay of landscape attributes across multiple spatial scales and site habitat attributes. Our study provides the first comprehensive insight into monarch use of conservation lands in Iowa, and predicted responses to important covariates may be useful for future conservation efforts.

Structured Genome-Wide Association Studies with Bayesian Hierarchical Variable Selection.

It becomes increasingly important in using genome-wide association studies (GWAS) to select important genetic information associated with qualitative or quantitative traits. Currently, the discovery of biological association among SNPs motivates various strategies to construct SNP-sets along the genome and to incorporate such set information into selection procedure for a higher selection power, while facilitating more biologically meaningful results. The aim of this paper is to propose a novel Bayesian framework for hierarchical variable selection at both SNP-set (group) level and SNP (within group) level. We overcome a key limitation of existing posterior updating scheme in most Bayesian variable selection methods by proposing a novel sampling scheme to explicitly accommodate the ultrahigh-dimensionality of genetic data. Specifically, by constructing an auxiliary variable selection model under SNP-set level, the new procedure utilizes the posterior samples of the auxiliary model to subsequently guide the posterior inference for the targeted hierarchical selection model. We apply the proposed method to a variety of simulation studies and show that our method is computationally efficient and achieves substantially better performance than competing approaches in both SNP-set and SNP selection. Applying the method to the Alzheimers Disease Neuroimaging Initiative (ADNI) data, we identify biologically meaningful genetic factors under several neuroimaging volumetric phenotypes. Our method is general and readily to be applied to a wide range of biomedical studies.

Landscape genetics of the foundational salt marsh plant species black needlerush (Juncus roemerianus Scheele) across the northeastern Gulf of Mexico

ContextCommon species important for ecosystem restoration stand to lose as much genetic diversity from anthropogenic habitat fragmentation and climate change as rare species, but are rarely studied. Salt marshes, valuable ecosystems in widespread decline due to human development, are dominated by the foundational plant species black needlerush (Juncus roemerianus Scheele) in the northeastern Gulf of Mexico.ObjectivesWe assessed genetic patterns in J. roemerianus by measuring genetic and genotypic diversity, and characterizing population structure. We examined population connectivity by delineating possible dispersal corridors, and identified landscape factors influencing population connectivity.MethodsA panel of 19 microsatellite markers was used to genotype 576 samples from ten sites across the northeastern Gulf of Mexico from the Grand Bay National Estuarine Research Reserve (NERR) to the Apalachicola NERR. Genetic distances (FST and Dch) were used in a least cost transect analysis (LCTA) within a hierarchical model selection framework.ResultsGenetic and genotypic diversity results were higher than expected based on life history literature, and samples structured into two large, admixed genetic clusters across the study area, indicating sexual reproduction may not be as rare as predicted in this clonal macrophyte. Digitized coastal transects buffered by 500 m may represent possible dispersal corridors, and developed land may significantly impede population connectivity in J. roemerianus.ConclusionsResults have important implications for coastal restoration and management that seek to preserve adaptive potential by sustaining natural levels of genetic diversity and conserving population connectivity. Our methodology could be applied to other common, widespread and understudied species.

A Bayesian Hierarchical Selection Model for Academic Growth With Missing Data

ABSTRACTUsing a sample of schools testing annually in grades 9–11 with a vertically linked series of assessments, a latent growth curve model is used to model test scores with student intercepts and slopes nested within school. Missed assessments can occur because of student mobility, student dropout, absenteeism, and other reasons. Missing data indicators are modeled using logistic regression, with grade 9 and potentially unobserved growth scores used as covariates. Under a hierarchical selection model, estimates of school effects on academic growth and missingness are obtained. The results from the selection model are compared to a model that ignores the missing data process.

Fitness Inference from Short-Read Data: Within-Host Evolution of a Reassortant H5N1 Influenza Virus.

We present a method to infer the role of selection acting during the within-host evolution of the influenza virus from short-read genome sequence data. Linkage disequilibrium between loci is accounted for by treating short-read sequences as noisy multilocus emissions from an underlying model of haplotype evolution. A hierarchical model-selection procedure is used to infer the underlying fitness landscape of the virus insofar as that landscape is explored by the viral population. In a first application of our method, we analyze data from an evolutionary experiment describing the growth of a reassortant H5N1 virus in ferrets. Across two sets of replica experiments we infer multiple alleles to be under selection, including variants associated with receptor binding specificity, glycosylation, and with the increased transmissibility of the virus. We identify epistasis as an important component of the within-host fitness landscape, and show that adaptation can proceed through multiple genetic pathways.

Publication bias in meta-analyses from the Cochrane Database of Systematic Reviews.

We used a Bayesian hierarchical selection model to study publication bias in 1106 meta-analyses from the Cochrane Database of Systematic Reviews comparing treatment with either placebo or no treatment. For meta-analyses of efficacy, we estimated the ratio of the probability of including statistically significant outcomes favoring treatment to the probability of including other outcomes. For meta-analyses of safety, we estimated the ratio of the probability of including results showing no evidence of adverse effects to the probability of including results demonstrating the presence of adverse effects. In the meta-analyses of efficacy, outcomes favoring treatment had on average a 27% (95% Credible Interval (CI): 18% to 36%) higher probability to be included than other outcomes. In the meta-analyses of safety, results showing no evidence of adverse effects were on average 78% (95% CI: 51% to 113%) more likely to be included than results demonstrating that adverse effects existed. In general, the amount of over-representation of findings favorable to treatment was larger in meta-analyses including older studies. In the largest study on publication bias in meta-analyses to date, we found evidence of publication bias in Cochrane systematic reviews. In general, publication bias is smaller in meta-analyses of more recent studies, indicating their better reliability and supporting the effectiveness of the measures used to reduce publication bias in clinical trials. Our results indicate the need to apply currently underutilized meta-analysis tools handling publication bias based on the statistical significance, especially when studies included in a meta-analysis are not recent.

General framework and model building in the class of Hidden Mixture Transition Distribution models

Modeling time series that present non-Gaussian features plays as central role in many fields, including finance, seismology, psychological, and life course studies. The Hidden Mixture Transition Distribution model is an answer to the complexity of such series. The observed heterogeneity can be induced by one or several latent factors, and each level of these factors is related to a different component of the observed process. The time series is then treated as a mixture and the relation between the components is governed by a Markovian latent transition process. This framework generalizes several specifications that appear separately in related literature. Both the expectation and the standard deviation of each component are allowed to be functions of the past of the process. The latent process can be of any order, and can be modeled using a discrete Mixture Transition Distribution. The effects of covariates at the visible and hidden levels are also investigated. One of the main difficulties lies in correctly specifying the structure of the model. Therefore, we propose a hierarchical model selection procedure that exploits the multilevel structure of our approach. Finally, we illustrate the model and the model selection procedure through a real application in social science.

Effects of urbanization on site occupancy and density of grassland birds in tallgrass prairie fragments

Tallgrass prairies are among the most threatened ecosystems in the world. Remaining prairies tend to be small and isolated and many are associated with urban and suburban landscapes. We asked how urbanization might impact the conservation value of tallgrass prairie fragments for grassland birds by comparing the densities and the probability of occurrence of Dickcissels (Spiza americana), Grasshopper Sparrows (Ammodramus savannarum), and Eastern Meadowlarks (Sturnella magna) across 28 grasslands surrounded by low, moderate, and high levels of urbanization. We employed a hierarchical model selection approach to ask how variables that describe the vegetation structure, size and shape of grasslands, and urbanization category might explain variation in density and occurrence over two breeding seasons. Occurrence of all three species was explained by a combination of vegetation and patch characteristics, though each species was influenced by different variables and only Eastern Meadowlark occurrence was explained by urbanization. Abundance of all three species was negatively impacted by urbanization, though vegetation variables were also prevalent in the best-supported models. We found no evidence that vegetation structure or other patch characteristics varied in a systematic way across urbanization categories. Although our results suggest that grassland bird density declines with urbanization, urban tallgrass prairies still retain conservation value for grassland birds because of the limited availability of tallgrass prairie habitat and the limited impact of urbanization on species occurrence.

Electoral entry and success of ethnic minority parties in central and eastern Europe: A hierarchical selection model

The paper examines determinants of electoral entry and success of ethnic minority parties in central and eastern Europe. The application of a hierarchical selection model shows that the strategic entry of minority parties depends on their expected electoral success due both to observed and unobserved factors. Drawing on formal models of electoral entry, the electoral success of new (or niche) parties is expected to be influenced by the costs of entry (determined by electoral thresholds) and the potential for electoral support. The latter depends on the reactions of political competitors and electoral demand, measured here as the size of ethnic groups and the saliency of ethnic issues. In line with these expectations, parties only run if they can expect electoral support sufficient to pass the electoral threshold. This finding would have been overlooked by a naïve model of electoral success which does not take self-selection into account.

Bayesian Normalized Gaussian Network and Hierarchical Model Selection Method

Abstract This paper presents a vaziational Bayes (VB) method for normalized Gaussian network, which is a mixture model of local experts. Based on the Bayesian framework, we introduce ameta-learning mechanism to optimize the prior distribution and the model structure. In order to search for the optimal model structure efficiently, we also develop a hierazchical model selection method. The performance of our method is evaluated by using function approximation problems and an system identification problem of a nonlinear dynamical system. Experimental results show that our Bayesian framework results in the reduction of generalization error and achieves better function approximation than existing methods within the fmite mixtures of experts family when the number of training data is fairly small.

Regression B‐spline smoothing in Bayesian disease mapping: with an application to patient safety surveillance

In the context of Bayesian disease mapping, recent literature presents generalized linear mixed models that engender spatial smoothing. The methods assume spatially varying random effects as a route to partially pooling data and 'borrowing strength' in small-area estimation. When spatiotemporal disease rates are available for sequential risk mapping of several time periods, the 'smoothing' issue may be explored by considering spatial smoothing, temporal smoothing and spatiotemporal interaction. In this paper, these considerations are motivated and explored through development of a Bayesian semiparametric disease mapping model framework which facilitates temporal smoothing of rates and relative risks via regression B-splines with mixed-effect representation of coefficients. Specifically, we develop spatial priors such as multivariate Gaussian Markov random fields and non-spatial priors such as unstructured multivariate Gaussian distributions and illustrate how time trends in small-area relative risks may be explored by splines which vary in either a spatially structured or unstructured manner. In particular, we show that with suitable prior specifications for the random effects ensemble, small-area relative risk trends may be fit by 'spatially varying' or randomly varying B-splines. A recently developed Bayesian hierarchical model selection criterion, the deviance information criterion, is used to assess the trade-off between goodness-of-fit and smoothness and to select the number of knots. The methodological development aims to provide reliable information about the patterns (both over space and time) of disease risks and to quantify uncertainty. The study offers a disease and health outcome surveillance methodology for flexible and efficient exploration and assessment of emerging risk trends and clustering. The methods are motivated and illustrated through a Bayesian analysis of adverse medical events (also known as iatrogenic injuries) among hospitalized elderly patients in British Columbia, Canada.

REVIEW OF PHENOTYPIC PLASTICITY AND HIERARCHICAL SELECTION IN CLONAL PLANTS

Phenotypic plasticity is the ability of a genotype to produce distinct phenotypes when exposed to different environments during its ontogeny. There is new strong evidence for the heritable nature of phenotypic plasticity, and variation in plasticity should thus be recognized as an integral component of evolution, but not as a factor that buffers and thereby constrains the action of selection. Clonal plants have both clonal modularity and organismic modularity. As a result, phenotypic selection in clonal plants is hierarchical. Modular hierarchy of clonal plants consists of three levels, the genet, ramet fragment and ramet levels. Modes of hierarchical selection and genotypic selection have been considered in studying natural selection in clonal plants. The hierarchical selection model treats each hierarchical level as a single level of phenotypic selection, and selection effects are determined by both the selective pressure on each level and ioteractions among levels. Net effects of multi-level phenotypic selection are transmitted finally to the succeeding generations through the ramet level. The genotypic selection model puts forward that clonal growth can lead to genetic variances between ramets within a genet. Then allele frequencies will change because of non-random mating within or among genets, leading to microevolution. In clonal plants, ramets may have the greatest phenotypic variation because it is the fundamental unit of sexual and asexual reproduction, whereas genets may have the lowest phenotypic variation because it is a relative stable unit. It is termed as the mode of hierarchical responses of phenotypic plasticity. The module of a plant has the greatest phenotypic plasticity, and phenotypic plasticity is not a whole-plant response but a property of module triggered by local environmental conditions. Therefore, the mode of hierarchical responses of phenotypic plasticity in clonal plants may be universal. If the mode indicates a response mode of adaptive plasticity of different `individuals' in modular hierarchy, we can predict that: 1) the effects of natural selection on clonal plants should be shown first at ramet level and will ultimately result in microevolution as predicted by both hierarchical selection and genotypic selection models and 2) the conservativeness in evolutionary changes may be greater in clonal plants than in non-clonal plants because of lower ability of sexual reproduction and high degree of physiological integration. The most promising directions for future research include areas such as those demonstrating the universality of hierarchical responses to natural selection in clonal plants and revealing the mechanisms of hierarchical responses.

Generalized Hierarchical Multivariate CAR Models for Areal Data

In the fields of medicine and public health, a common application of areal data models is the study of geographical patterns of disease. When we have several measurements recorded at each spatial location (for example, information on p>/= 2 diseases from the same population groups or regions), we need to consider multivariate areal data models in order to handle the dependence among the multivariate components as well as the spatial dependence between sites. In this article, we propose a flexible new class of generalized multivariate conditionally autoregressive (GMCAR) models for areal data, and show how it enriches the MCAR class. Our approach differs from earlier ones in that it directly specifies the joint distribution for a multivariate Markov random field (MRF) through the specification of simpler conditional and marginal models. This in turn leads to a significant reduction in the computational burden in hierarchical spatial random effect modeling, where posterior summaries are computed using Markov chain Monte Carlo (MCMC). We compare our approach with existing MCAR models in the literature via simulation, using average mean square error (AMSE) and a convenient hierarchical model selection criterion, the deviance information criterion (DIC; Spiegelhalter et al., 2002, Journal of the Royal Statistical Society, Series B64, 583-639). Finally, we offer a real-data application of our proposed GMCAR approach that models lung and esophagus cancer death rates during 1991-1998 in Minnesota counties.

Fatores de risco para anemia em crianças de 6 a 12 meses no Brasil

To estimate the prevalence of anemia and to determine associated risk factors among infants receiving routine health care in public clinics in Brazil. This cross-sectional study included 2,715 infants between 6 and 12 months old in 12 cities, in all five of the geographic regions of Brazil. Information regarding the child and its feeding habits was obtained from the mother or other caregiver, using a questionnaire. Nutritional status was determined based on height and weight measurements. The hemoglobin concentration was measured using the HemoCue portable hemoglobinometer. Anemia was defined as hemoglobin < 11 g/dL. The infants' eating habits were assessed based on what they were eating around the time of the questionnaire interviews. The association between anemia and the different variables was evaluated through bivariate analysis, followed by multiple logistic regression using a hierarchical selection model. The prevalence of anemia for the entire group was 65.4%. Multiple regression analysis identified the following risk factors for anemia: living in the Southeastern Region of Brazil (odds ratio (OR) = 1.57, 95% confidence interval (95% CI) = 1.25- 1.99), maternal age < 20 years (OR = 1.58, 95% CI = 1.21-2.07), birthweight < 2,500 g (OR = 2.09, 95% CI = 1.48-2.95), not being breast-fed (OR = 1.28, 95% CI = 1.02-1.61), receiving both breast milk and other foods (OR = 1.40, 95% CI = 1.10-1.78), and male gender (OR = 1.24, 95% CI = 1.06-1.46). The high proportion of anemic children indicates the need to emphasize, in prenatal and infant health programs, intervention measures for anemia control. Our results could guide these measures, focusing on the groups at greatest risk, such as low birthweight babies and the children of adolescent mothers.

変分法的ベイズ推定法に基づく正規化ガウス関数ネットワークと階層的モデル選択法

変分法的ベイズ推定法に基づく正規化ガウス関数ネットワークと階層的モデル選択法