Bayesian chain graph models to characterize microbe-environment dynamics.

One of the most characteristic features of biological molecular networks is that the network structure itself changes, depending on the cellular environment. Indeed, activated molecules show a variety of responses to distinctive cell conditions, and subsequently the network structures of active molecules also change. Here we present an approach to trace the network structure changes by using the graphical chain model developed from the gene expression data. The previous procedure for applying the graphical chain model to the expression profiles of a limited number of genes has been improved to analyze the entire set of genes. Furthermore, the chain model has been rearranged according to the association strength, and was scrutinized to identify the candidates of essential gene-gene relationships for the network changes, by using the path consistency algorithm. The improved procedure was applied to the expression profiles of 8,427 genes, which were measured in two distinctive stages of liver cancer progression. As a result, the chain model of the 18 gene cluster relationships with strong associations was inferred, in which the coordination of clusters was described in the cell stage progression, and the gene-gene relationships between known cancer-related genes causing the progression were further refined. Thus, the present procedure is a useful method to model the network structure changes in the cell stage progression, and to clarify the gene candidates for the progression.

High correlations between MRI brain volume measurements based on NeuroQuant® and FreeSurfer

In Bayesian Network Regression models, networks are considered the predictors of continuous responses. These models have been successfully used in brain research to identify regions in the brain that are associated with specific human traits, yet their potential to elucidate microbial drivers in biological phenotypes for microbiome research remains unknown. In particular, microbial networks are challenging due to their high dimension and high sparsity compared to brain networks. Furthermore, unlike in brain connectome research, in microbiome research, it is usually expected that the presence of microbes has an effect on the response (main effects), not just the interactions. Here, we develop the first thorough investigation of whether Bayesian Network Regression models are suitable for microbial datasets on a variety of synthetic and real data under diverse biological scenarios. We test whether the Bayesian Network Regression model that accounts only for interaction effects (edges in the network) is able to identify key drivers (microbes) in phenotypic variability. We show that this model is indeed able to identify influential nodes and edges in the microbial networks that drive changes in the phenotype for most biological settings, but we also identify scenarios where this method performs poorly which allows us to provide practical advice for domain scientists aiming to apply these tools to their datasets. BNR models provide a framework for microbiome researchers to identify connections between microbes and measured phenotypes. We allow the use of this statistical model by providing an easy-to-use implementation which is publicly available Julia package at https://github.com/solislemuslab/BayesianNetworkRegression.jl.

Protein secondary structure prediction is an important step to understanding protein tertiary structure. Recent studies indicate that the correlation between neighboring secondary structures are beneficial to improve prediction performance. In this paper, we propose a new large margin approach for protein secondary structure prediction, which consider the problem as a sequence labeling problem like probabilistic graphical models. It doesn’t only make full use of the correlation between neighboring secondary structures like graphical chain models, but also shares the key advantages of other SVM-based methods, i.e. learning non-linear discriminate via kernel functions. The experimental results on datasets: CB513 and RS126 show that our algorithm outperforms other state-of-the-art methods.

Hypothermia is a major cause of mortality in neonatal piglets. However, there are considerable individual differences in the successful recovery from postnatal hypothermia in the common farrowing environment, and so far the causes and interactions of causes have not been studied in detail. Using 635 crossbred neonatal piglets, the aim of this study was to identify the links among different physiological and behavioral measures and their connections to the ability of piglets to overcome initial postnatal hypothermia, with rectal temperature at 2 h as the response variable. The data included birth weight, hypoxia at birth (viability score and lactate in umbilical cord blood), latency to first udder contact and first suckle, scans of individual piglet position during the first 2 h after birth, and rectal temperature at birth and 2 h postpartum. A graphical chain model was used to analyze data. The statistical variables were divided into blocks according to level (design and litter) and chronological order (prenatal, birth, perinatal, and thermoregulatory success at 2 h) before applying the graphical model to the data. Bayesian information criteria (BIC) was used for model selection. The BIC relates to maximum likelihood, but introduces an additional penalty term for the number of variables. The strength of an association between 2 variables is reported as the increase in BIC (BICinc) due to removing the link. Results indicate that at 2 h, 22.1% of the piglets had a rectal temperature below 37 °C. Out of the 16 variables included in the model, only 3 had direct links to the response variable of rectal temperature at 2 h. There was a positive relationship between rectal temperature at 2 h and birth weight (BICinc=26), and between being observed more often by the udder as opposed to alone during both the first (BICinc=8) and second hours (BICinc=19) after birth. Lighter piglets and piglets that had experienced hypoxia took longer to achieve first suckle, which in turn affected where the piglet positioned itself during the first and second hours after birth. Variables related to the birth process had no direct connection to thermoregulatory success, but were additive in the explanation of piglet behavior. The rectal temperature of individual piglets at 2 h depends largely on piglet birth weight and on piglet position in relation to sow and littermates. Birth weight is the most important single factor in successful recovery from postnatal hypothermia.

On estimation in the reduced-rank regression with a large number of responses and predictors

Gaussian Process (GP) regression has seen widespread use in robotics due to its generality, simplicity of use, and the utility of Bayesian predictions. The predominant implementation of GP regression is a nonparametric kernel-based approach, as it enables fitting of arbitrary nonlinear functions. However, this approach suffers from two main drawbacks: (1) it is computationally inefficient, as computation scales poorly with the number of samples; and (2) it can be data inefficient, as encoding prior knowledge that can aid the model through the choice of kernel and associated hyperparameters is often challenging and unintuitive. In this work, we propose ALPaCA, an algorithm for efficient Bayesian regression which addresses these issues. ALPaCA uses a dataset of sample functions to learn a domain-specific, finite-dimensional feature encoding, as well as a prior over the associated weights, such that Bayesian linear regression in this feature space yields accurate online predictions of the posterior predictive density. These features are neural networks, which are trained via a meta-learning (or “learning-to-learn”) approach. ALPaCA extracts all prior information directly from the dataset, rather than restricting prior information to the choice of kernel hyperparameters. Furthermore, by operating in the weight space, it substantially reduces sample complexity. We investigate the performance of ALPaCA on two simple regression problems, two simulated robotic systems, and on a lane-change driving task performed by humans. We find our approach outperforms kernel-based GP regression, as well as state of the art meta-learning approaches, thereby providing a promising plug-in tool for many regression tasks in robotics where scalability and data-efficiency are important.

The purpose of this paper is to analyse the herding behavior of eight stock sectors of the Stock Exchange of Thailand consisting of Banking (BANK), Commerce (COM), Communications (COMUN), Energy and Utilities (ENERG), Food and Beverage (FOOD), Personal Products and Pharmaceuticals (PERS), Petrochemicals and Chemicals (PETRO), and Property Development (PROP), for the period from January of 2014 to May of 2018. Conventionally, the model used for finding the herding behavior is linear regression, which is based on the mean of the distribution. However, the result obtained from the linear mean regression model may not be consistent with the heterogeneity of investor behavior. Thus, in this study, we employ Bayesian quantile regression to empirically estimate the daily stock returns. The innovation of this paper is to examine the data conditional on different quantiles and test the behavioral relation between stock returns and market movements with different quantile distributions. The results show that five out of the eight sectors have herding behavior at quantile 0.25 and the remaining three sectors have no herding behavior.

It is thought that physical health conditions start at a young age in people with profound intellectual and multiple disabilities (PIMD). Knowledge regarding the prevalence, associations and development of these physical health conditions could be used for purposes of prevention as well as appropriate care and support but is currently lacking. The aim of this study is to gain insight into the prevalence of physical health conditions and associations between these conditions in young children with PIMD. The study used cross-sectional data related to the physical health conditions of children with PIMD (n= 51, aged between 12 and 61 months). Data were collected in Belgium and in the Netherlands through a checklist filled in by primary caregiver(s). Physical health conditions were classified into categories by the 10th revision of the International Classification of Diseases and Related Health Problems (ICD-10) system. The number of physical health conditions and associations between them were analysed. The analysis focused on prevalence rates and associations represented by odds ratios (p< 0.05). A graphical model was estimated to represent dependencies and conditional dependencies between physical health conditions. We found a mean of 3.8 (range 1-8, SD 1.9) physical health conditions per child. Most of the physical health conditions were found in the ICD-10 chapter 'Nervous System', with hypotonia as the most frequent at 70.6%. Five significant large associations were found between spasticity-contractures (OR 9.54); circulatory system-contractures (OR 7.50); scoliosis-contractures (OR 10.25); hearing impairments-skin problems (OR 58.20) and obstipation-hypotonia (OR 19.98). This study shows that at a young age, multiple physical health conditions are present in children with PIMD. In addition, we found five associations between physical health conditions.

In this paper, a hybrid modeling framework integrating multiple linear regression, Bayesian regression and decision tree is proposed for target variable prediction. First, the quantitative prediction of target variables is realized by constructing a multivariate linear regression model with a five-dimensional feature space, and the model is evaluated for performance using F-test and R² value to verify the significance and explanatory power of the combination of independent variables. Secondly, Bayesian regression method is introduced to deal with the binary prediction task, which improves the uncertainty quantification ability through probabilistic modeling and optimizes the feature screening process by combining with the decision tree algorithm, and finally outputs the target breakthrough probability distribution. Finally, the dominant feature dimension is extended to seven dimensions through feature engineering, and the improved model significantly improves the prediction accuracy while maintaining the computational efficiency, which verifies the key role of feature selection on the model performance. The experimental results show that the hybrid model framework has good generalization ability and interpretability in complex prediction scenarios.

&lt;div class="section abstract"&gt;&lt;div class="htmlview paragraph"&gt;It is an important factor in electric vehicles to show customers how much they can drive with the energy of the remaining battery. If the remaining mileage is not accurate, electric vehicle drivers will have no choice but have to feel anxious about the mileage. Additionally, the potential customers have range anxiety when they consider Electric Vehicles. If the remaining mileage to drive is wrong, drivers may not be able to get to the charging station and may not be able to drive because the battery runs out. It is important to show the remaining available driving range exactly for drivers. The previous study proposed an advanced model by predicting the remaining mileage based on actual driving data and based on reflecting the pattern of customers who drive regularly. The Bayesian linear regression model was right model in previous study. In addition, in order to improve performance, the driver's regular driving pattern is recognized in advance before driving and it is reflected in the remaining driving mileage model with Bayesian regression. It could be seen that the performance of the model in previous study was improved 10% better compared to the remaining driving mileage existed in vehicle function. The purpose of this study confirms the robust performance with vast data of the more vehicles in long mileage. The big data are from that almost vehicles took drive for 10k mile range. The vehicles had various experience in location, driving time rage, mileages, driving pattern, charge pattern, and so on. Moreover, the personalized model gets improved in big data to get higher accuracy. This study proposes personalized algorithm enable to embed in vehicles.&lt;/div&gt;&lt;/div&gt;

Tuberculosis (TB) continues to be a major public health concern in India, particularly in socioeconomically disadvantaged regions where nutritional deficiencies are widespread. Madhya Pradesh, one of the high-burden states, exhibits a concerning overlap between elevated TB incidence and chronic malnutrition among women and children. This study aims to quantify the relationship between nutritional indicators and TB prevalence across ten selected districts in Madhya Pradesh using Bayesian regression analysis. Secondary data were sourced from the NIKSHAY portal (2015–2022) for TB cases and from NFHS-5 and the State Nutrition Profile (2019–2021) for malnutrition indicators, including anemia, stunting, and underweight status. Descriptive statistics and correlation analyses revealed strong associations between TB prevalence and nutritional deficits, particularly anemia in children and underweight status in women. Bayesian linear regression was employed to model the predictive strength of five key nutritional variables. The best-fitting model explained 95% of the variance in TB prevalence (R² = 0.950), with high posterior inclusion probabilities and Bayes Factors supporting the relevance of each predictor. Notably, anemia in children emerged as the most influential variable, followed by underweight women, while stunting and underweight children showed complex or inverse associations.These findings reinforce the hypothesis that malnutrition significantly contributes to TB vulnerability and underscore the importance of integrating nutritional interventions into TB control programs. By applying Bayesian methods, this study offers a robust statistical framework for identifying high-risk populations and guiding targeted public health strategies. The results have implications for policy formulation, resource allocation, and the design of integrated health programs in Madhya Pradesh and similar high-burden regions.

A continuum-particle coupling approach based on Bayesian regression, the micro–macro-surrogate-sparse method, is presented. In contrast to the domain-decomposition technique, we use Bayesian linear regression to infer constitutive and boundary-condition corrections from limited direct simulation Monte Carlo (DSMC) data. These corrections are incorporated into a computational fluid dynamics (CFD) framework to calculate macroscopic properties. This work builds upon the methodology proposed in Tatsios et al. [“A DSMC-CFD coupling method using surrogate modelling for low-speed rarefied gas flows,” J. Comput. Phys. 520, 113500 (2025)], with greater flexibility, autonomy and generality; this is showcased through application to a two-dimensional (spatio-temporal) rarefied-gas problem. In the coupled model, the DSMC does not directly calculate the flow in non-equilibrium regions; instead, it is used only to find corrections required by the Navier–Stokes equations. As a demonstration of the benefits of the methodology, unsteady rarefied force-driven Poiseuille flow is studied and the role of three key non-dimensional parameters explored. The proposed solver is validated for a range of flow conditions by comparing the results to expensive benchmark DSMC simulations. It is found that the proposed model accurately predicts non-equilibrium effects and is effective in capturing several phenomena, including the well-known Richardson effect at high oscillation frequencies and a non-linear variation of slip velocity at high forcing amplitudes. For initial proof-of-concept testing, the corrections are obtained using DSMC data from the entire domain. After this, to test ability to make predictions with partial data, only DSMC data obtained close to walls are used to train the model. Error analyses, computational cost comparisons, and the effect of noise on the results are described. In all cases, the proposed model exhibits higher accuracy than the DSMC data used to train the hybrid scheme, while far outperforming the benchmark DSMC data in terms of computational efficiency.

The understanding of co-movements, dependence, and influence between variables of interest is key in many applications. Broadly speaking such understanding can lead to better predictions and decision making in many settings. We propose Quantile Graphical Models (QGMs) to characterize prediction and conditional independence relationships within a set of random variables of interest. Although those models are of interest in a variety of applications, we draw our motivation and contribute to the financial risk management literature. Importantly, the proposed framework is intended to be applied to non-Gaussian settings, which are ubiquitous in many real applications, and to handle a large number of variables and conditioning events. We propose two distinct QGMs. First, Condition Independence Quantile Graphical Models (CIQGMs) characterize conditional independence at each quantile index revealing the distributional dependence structure. Second, Prediction Quantile Graphical Models (PQGMs) characterize the best linear predictor under asymmetric loss functions. A key difference between those models is the (non-vanishing) misspecification between the best linear predictor and the conditional quantile functions. We also propose estimators for those QGMs. Due to high-dimensionality, the two distinct QGMs require different estimators. The estimators are based on high-dimensional techniques including (a continuum of) `1-penalized quantile regressions (and low biased equations), which allow us to handle the potential large number of variables. We build upon a recent literature to obtain new results for valid choice of the penalty parameters, rates of convergence, and confidence regions that are simultaneously valid. We illustrate how to use QGMs to quantify tail interdependence (instead of mean dependence) between a large set of variables which is relevant in applications concerning with extreme events. We show that the associated tail risk network can be used for measuring systemic risk contributions. We also apply the framework to study international financial contagion and the impact of market downside movement on the dependence structure of assets' returns.

Many population-based surveys have binary responses from a large number of individuals in each household within small areas. One example is the Nepal Living Standards Survey (NLSS II), in which health status binary data (good versus poor) for each individual from sampled households (sub-areas) are available in the sampled wards (small areas). To make an inference for the finite population proportion of individuals in each household, we use the sub-area logistic regression model with reliable auxiliary information. The contribution of this model is twofold. First, we extend an area-level model to a sub-area level model. Second, because there are numerous sub-areas, standard Markov chain Monte Carlo (MCMC) methods to find the joint posterior density are very time-consuming. Therefore, we provide a sampling-based method, the integrated nested normal approximation (INNA), which permits fast computation. Our main goal is to describe this hierarchical Bayesian logistic regression model and to show that the computation is much faster than the exact MCMC method and also reasonably accurate. The performance of our method is studied by using NLSS II data. Our model can borrow strength from both areas and sub-areas to obtain more efficient and precise estimates. The hierarchical structure of our model captures the variation in the binary data reasonably well.