OBJECTIVE Hemoglobin A1c (A1C) is used in assessment of patients for elective surgeries because hyperglycemia increases risk adverse events. However, the interplay A1C, glucose, and surgical outcomes remains unclarified, with often only two these three factors considered simultaneously. We assessed association preoperative A1C perioperative glucose control their relationship 30-day mortality. RESEARCH DESIGN AND METHODS Retrospective analysis on 431,480 within Duke University Health System...

Background: Precision medicine (PM) programs typically use broad consent. This approach requires maintenance of the social license and public trust. The ultimate success PM will thus likely be contingent upon understanding expectations about data sharing establishing appropriate governance structures. There is a lack on attitudes towards in Asia. Methods: aim research was to measure priorities preferences Singaporeans for health-related PM. We used adaptive choice-based conjoint analysis...

Gaussian graphical models can capture complex dependency structures among variables. For such models, Bayesian inference is attractive as it provides principled ways to incorporate prior information and quantify uncertainty through the posterior distribution. However, computation under conjugate G-Wishart distribution on precision matrix expensive for general non-decomposable graphs. We therefore propose a new Markov chain Monte Carlo (MCMC) method named weighted proposal algorithm (WWA)....

Given a large clinical database of longitudinal patient information including many covariates, it is computationally prohibitive to consider all types interdependence between variables interest. This challenge motivates the use mutual (MI), statistical summary data with appealing properties that make suitable alternative or addition correlation for identifying relationships in data. MI: (i) captures dependence, both linear and nonlinear, (ii) zero only when random are independent, (iii)...

Several applications involving counts present a large proportion of zeros (excess-of-zeros data). A popular model for such data is the hurdle model, which explicitly models probability zero count, while assuming sampling distribution on positive integers. We consider from multiple count processes. In this context, it interest to study patterns and cluster subjects accordingly. introduce novel Bayesian approach multiple, possibly related, zero-inflated propose joint counts, specifying each...

Probabilistically quantifying uncertainty in parameters, predictions and decisions is a crucial component of broad scientific engineering applications. This however difficult if the number parameters far exceeds sample size. Although there are currently many methods which have guarantees for problems characterized by large random matrices, often gap between theory practice when it comes to measures statistical significance matrices encountered real-world paper proposes scalable framework...

In many contexts, there is interest in selecting the most important variables from a very large collection, commonly referred to as support recovery or variable, feature subset selection. There an enormous literature proposing rich variety of algorithms. scientific applications, it crucial importance quantify uncertainty variable selection, providing measures statistical significance for each variable. The overwhelming majority algorithms fail produce such measures. This has led focus on...

Maternal depression and anxiety through pregnancy have lasting societal impacts. It is thus crucial to understand the trajectories of its progression from preconception postnatal period, risk factors associated with it. Within Bayesian framework, we propose jointly model seven outcomes, which two are physiological five non-physiological indicators maternal over time. We former by a Gaussian process latter an autoregressive model, while imposing multidimensional Dirichlet prior on...

Graphical models provide a powerful methodology for learning the conditional independence structure in multivariate data. Inference is often focused on estimating individual edges latent graph. Nonetheless, there increasing interest inferring more complex structures, such as communities, multiple reasons, including effective information retrieval and better interpretability. Stochastic blockmodels offer tool to detect network. We thus propose exploit advances random graph theory embed them...

Abstract The objective of this study is to obtain accurate and timely Account Receivables forecasts that can feed into the Cash Flow Forecasting (CFF) for a generic wholesales scenario in supply chain company. main components CFF are (AR), Payables (AP) Working Capital (WC). focus work on AR, particular how they contribute overall cash flow forecasting. prediction AR based predicted payment date each invoice, which be obtained from time given issue date. In context, we propose use discrete...

We propose an adaptive importance sampling scheme for Gaussian approximations of intractable posteriors. Optimization-based like variational inference can be too inaccurate while existing Monte Carlo methods slow. Therefore, we a hybrid where, at each iteration, the effective sample size guaranteed fixed computational cost by interpolating between natural-gradient and sampling. The amount damping in updates adapts to posterior guarantees size. Gaussianity enables use Stein's lemma obtain...

Time-to-event data are often recorded on a discrete scale with multiple, competing risks as potential causes for the event. In this context, application of continuous survival analysis methods single risk suffers from biased estimation. Therefore, we propose multivariate Bernoulli detector times involving change point model cause-specific baseline hazards. Through prior number points and their location, impose dependence between across risks, well allowing data-driven learning number. Then,...

Abstract Gaussian graphical models are useful tools for conditional independence structure inference of multivariate random variables. Unfortunately, Bayesian latent graph structures is challenging due to exponential growth $\mathcal{G}_n$ , the set all graphs in n vertices. One approach that has been proposed tackle this problem limit search subsets . In paper we study vector subspaces with cycle space $\mathcal{C}_n$ as main example. We propose a novel prior on based linear combinations...

The number of recurrent events before a terminating event is often interest. For instance, death terminates an individual's process rehospitalizations and the important indicator economic cost. We propose model in which recurrences termination random variable interest, enabling inference prediction on it. Then, conditionally this number, we specify joint distribution for recurrence survival. This novel conditional approach induces dependence between survival, present, instance due to frailty...

Although there is a rich literature on methods for assessing the impact of functional predictors, focus has been approaches dimension reduction that do not suit certain applications. Examples standard include linear models, principal components regression and cluster-based approaches, such as latent trajectory analysis. This article motivated by applications in which dynamics predictor, across times when value relatively extreme, are particularly informative about response. For example,...

Posterior computation for high-dimensional data with many parameters can be challenging. This article focuses on a new method approximating posterior distributions of low- to moderate-dimensional parameter in the presence or otherwise computationally challenging nuisance parameter. The focus is regression models and key idea separate likelihood into two components through rotation. One component involves only parameters, which then integrated out using novel type Gaussian approximation. We...

Time-to-event data are often recorded on a discrete scale with multiple, competing risks as potential causes for the event. In this context, application of continuous survival analysis methods single risk suffer from biased estimation. Therefore, we propose Multivariate Bernoulli detector times involving multivariate change point model cause-specific baseline hazards. Through prior number points and their location, impose dependence between across risks, well allowing data-driven learning...