Article
Automation & Control Systems
Trambak Banerjee, Qiang Liu, Gourab Mukherjee, Wenguang Sun
Summary: The NEB framework is a nonparametric empirical Bayes method for estimation in the discrete linear exponential family, with strong asymptotic properties and flexibility to incorporate various constraints, providing a unified approach to estimation. Comprehensive simulation studies and analysis of real data examples demonstrate the superiority of the NEB estimator over competing methods.
JOURNAL OF MACHINE LEARNING RESEARCH
(2021)
Article
Economics
Timothy B. Armstrong, Michal Kolesar, Mikkel Plagborg-Moller
Summary: This study constructs robust empirical Bayes confidence intervals for normal means problem, which are not affected by the means distribution and have an average coverage guarantee.
Article
Biochemical Research Methods
Javier E. Flores, Lisa M. Bramer, David J. Degnan, Vanessa L. Paurus, Yuri E. Corilo, Chaevien S. Clendinen
Summary: Gas chromatography-mass spectrometry (GC-MS) is an analytic method used to identify small molecules (e.g., metabolites) in metabolomics research. However, there is a risk of false identification or discovery that is not quantified. To address this, researchers propose a model-based framework for estimating the false discovery rate (FDR) among a set of identifications.
JOURNAL OF THE AMERICAN SOCIETY FOR MASS SPECTROMETRY
(2023)
Article
Anesthesiology
David Sidebotham, C. Jake Barlow, Janet Martin, Philip M. Jones
Summary: Randomized controlled trials are important for quantifying the effectiveness of medical interventions. This article explores the relationship between treatment effects and statistical significance based on P values and confidence intervals. It introduces Bayesian methods as an alternative to frequentist hypothesis testing and proposes simplified Bayesian metrics as an interim solution to the significance problem in medical research.
CANADIAN JOURNAL OF ANESTHESIA-JOURNAL CANADIEN D ANESTHESIE
(2023)
Article
Statistics & Probability
Sihai Dave Zhao, William Biscarri
Summary: This article proposes an alternative paradigm based on regression modeling to address the challenge of incorporating structural information in simultaneous estimation of multiple parameters. The approach is able to effectively handle denoising gene expression data.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Biology
Yi Li, Yaning Yang, Xu Steven Xu, Min Yuan
Summary: In this paper, we investigate the application of the empirical Bayes method in mixed-effects models with multiple covariates. By deriving the shrinkage factor and variance-covariance matrix, we propose a method to correct the empirical Bayes estimates and test statistics. The validity of the proposed method is demonstrated through theoretical derivations, simulation studies, and a real data application.
COMPUTATIONAL BIOLOGY AND CHEMISTRY
(2022)
Article
Biochemical Research Methods
Dominik Madej, Long Wu, Henry Lam
Summary: This study introduces a new method called CDD for FDR estimation, utilizing a fixed empirical null score distribution. Benchmarking CDD against decoy-based PeptideProphet showed similar accuracy and stability in retrieving correct PSMs. This finding highlights the potential of Big Data approaches for statistical analysis in proteomics and questions the necessity of dataset-specific target-decoy searches.
JOURNAL OF PROTEOME RESEARCH
(2022)
Article
Neurosciences
Seok-Oh Jeong, Jiyoung Kang, Chongwon Pae, Jinseok Eo, Sung Min Park, Junho Son, Hae-Jeong Park
Summary: The pairwise maximum entropy model (pMEM) has gained attention for exploring non-linear characteristics in brain state dynamics observed in resting-state functional magnetic resonance imaging (rsfMRI). An empirical Bayes estimation of individual pMEM using the variational expectation-maximization algorithm (VEM-MEM) is proposed, which can reliably estimate individual model parameters by incorporating group information as prior. The study shows that the proposed method is effective in characterizing dynamic properties of individuals and is more sensitive to group differences and behavior scores, compared to conventional functional connectivity methods.
Article
Statistics & Probability
Nikolaos Ignatiadis, Stefan Wager
Summary: In this paper, the authors develop flexible and practical confidence intervals for empirical Bayes estimands. These intervals have asymptotic frequentist coverage, even when the estimands are only partially identified or when empirical Bayes point estimates converge very slowly.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Statistics & Probability
Nicholas C. Henderson, Ravi Varadhan, Thomas A. Louis
Summary: This article introduces a new method of regression weights for estimating small domain means under potential misspecification of the global regression model. The method combines model-based weights and observed weights to preserve robustness and advantages.
JOURNAL OF THE AMERICAN STATISTICAL ASSOCIATION
(2022)
Article
Biochemical Research Methods
Sangjeong Lee, Heejin Park, Hyunwoo Kim
Summary: Accurately estimating the false discovery rate (FDR) is crucial in proteomics. The traditional target-decoy strategy (TDS) assumes the probabilities of spectra matching target or decoy peptides are identical, but in reality, they are not. Therefore, we propose cTDS, which estimates the FDR more accurately by considering the probability of spectra being incorrectly identified as target or decoy peptides.
BMC BIOINFORMATICS
(2022)
Article
Computer Science, Artificial Intelligence
Zhanhao Peng, Qing Zhou
Summary: This work proposes a hierarchical model and develops a novel empirical Bayes estimate for the connectivity matrix of a stochastic block model to approximate the graphon function, introducing a new model selection criterion for choosing the number of communities. Numerical results on extensive simulations and two well-annotated social networks demonstrate the superiority of the approach in terms of parameter estimation and model selection.
PEERJ COMPUTER SCIENCE
(2022)
Article
Mathematical & Computational Biology
Srijata Samanta, Joseph Antonelli
Summary: The analysis of environmental mixtures is becoming increasingly important in environmental epidemiology. However, current approaches lack the ability to control error rates and identify exposures and interactions. This study proposes two novel methods that simultaneously estimate the overall mixture effect, provide valid inference, and control the false discovery rate to identify important exposures and interactions. The results show increased power in detecting weak effects of environmental exposures.
Article
Physics, Multidisciplinary
Salima Helali, Afif Masmoudi, Yousri Slaoui
Summary: This paper focuses on addressing the boundary problem in probability density functions with bounded support. The use of mixtures of beta densities has led to various methods for estimating densities of data with compact support. Among these methods, the Bernstein polynomials have been found to improve the accuracy of edge estimation for density functions. This paper proposes a shrinkage method that combines Bernstein polynomials and a finite Gaussian mixture model to construct a semi-parametric density estimator, which enhances the approximation accuracy at the boundaries. The proposed approach is evaluated through simulation studies and real data sets, investigating its convergence properties and asymptotic normality.
Article
Social Sciences, Interdisciplinary
Jin Yuan, Xianghui Yuan
Summary: Covariance matrix estimation is important in portfolio analysis and risk management. This paper proposes a method to enhance covariance matrix estimation by utilizing additional data prior and factor models. Theoretical results show that the proposed estimators outperform existing methods. Numerical examples and simulations validate the effectiveness of the proposed methods.