Search

We consider a new approach for estimating non‐Gaussian undirected graphical models. Specifically, we model continuous data from a class of multivariate skewed distributions, whose conditional dependence structure depends on both a precision matrix and a shape vector. To estimate the graph, we propose a novel estimation method based on nodewise regression: we first fit a linear model, and then fit a one component projection pursuit regression model to the residuals obtained from the linear model, and finally threshold appropriate quantities. Theoretically, we establish error bounds for each nodewise regression and prove the consistency of the estimated graph when the number of variables diverges with the sample size. Simulation results demonstrate the strong finite sample performance of our new method over existing methods for estimating Gaussian and non‐Gaussian graphical models. Finally, we demonstrate an application of the proposed method on observations of physicochemical properties of wine. [ABSTRACT FROM AUTHOR]

Summary: This article re‐examines the F‐test based on linear combinations of the responses, or FLC test, for testing random effects in linear mixed models. In current statistical practice, the FLC test is underused and we argue that it should be reconsidered as a valuable method for use with linear mixed models. We present a new, more general derivation of the FLC test which applies to a broad class of linear mixed models where the random effects can be correlated. We highlight three advantages of the FLC test that are often overlooked in modern applications of linear mixed models, namely its computation speed, its generality, and its exactness as a test. Empirical studies provide new insight into the finite sample performance of the FLC test, identifying cases where it is competitive or even outperforms modern methods in terms of power, as well as settings in which it performs worse than simulation‐based methods for testing random effects. In all circumstances, the FLC test is faster to compute. The F‐test based on linear combinations of the responses is re examined as a general, fast, and exact method for testing random effects in mixed models. [ABSTRACT FROM AUTHOR]

Motivation: Gut microbiota can be classified at multiple taxonomy levels. Strategies to use changes in microbiota composition to effect health improvements require knowing at which taxonomy level interventions should be aimed. Identifying these important levels is difficult, however, because most statistical methods only consider when the microbiota are classified at one taxonomy level, not multiple.Results: Using L1 and L2 regularizations, we developed a new variable selection method that identifies important features at multiple taxonomy levels. The regularization parameters are chosen by a new, data-adaptive, repeated cross-validation approach, which performed well. In simulation studies, our method outperformed competing methods: it more often selected significant variables, and had small false discovery rates and acceptable false-positive rates. Applying our method to gut microbiota data, we found which taxonomic levels were most altered by specific interventions or physiological status.Availability: The new approach is implemented in an R package, which is freely available from the corresponding author.Contact: tpgarcia@srph.tamhsc.eduSupplementary information: Supplementary data are available at Bioinformatics online. [ABSTRACT FROM AUTHOR]

Variational Bayes (VB) estimation is a fast alternative to Markov Chain Monte Carlo for performing approximate Baesian inference. This procedure can be an efficient and effective means of analyzing large datasets. However, VB estimation is often criticised, typically on empirical grounds, for being unable to produce valid statistical inferences. In this article we refute this criticism for one of the simplest models where Bayesian inference is not analytically tractable, that is, the Bayesian linear model (for a particular choice of priors). We prove that under mild regularity conditions, VB based estimators enjoy some desirable frequentist properties such as consistency and can be used to obtain asymptotically valid standard errors. In addition to these results we introduce two VB information criteria: the variational Akaike information criterion and the variational Bayesian information criterion. We show that variational Akaike information criterion is asymptotically equivalent to the frequentist Akaike information criterion and that the variational Bayesian information criterion is first order equivalent to the Bayesian information criterion in linear regression. These results motivate the potential use of the variational information criteria for more complex models. We support our theoretical results with numerical examples. [ABSTRACT FROM AUTHOR]

We revisit Hawkins' (Comput Stat 9(3):233-247, ) algorithm for fitting monotonic polynomials and discuss some practical issues that we encountered using this algorithm, for example when fitting high degree polynomials or situations with a sparse design matrix but multiple observations per $$x$$-value. As an alternative, we describe a new approach to fitting monotone polynomials to data, based on different characterisations of monotone polynomials and using a Levenberg-Marquardt type algorithm. We consider different parameterisations, examine effective starting values for the non-linear algorithms, and discuss some limitations. We illustrate our methodology with examples of simulated and real world data. All algorithms discussed in this paper are available in the R Development Core Team (A language and environment for statistical computing, R Foundation for Statistical Computing, Vienna, ) package MonoPoly. [ABSTRACT FROM AUTHOR]

Objective The aim was to develop a new model to predict the outcome at the end of the 1st trimester after a single visit to the early pregnancy unit ( EPU). Methods Prospective observational study in the EPU at Nepean Hospital, between November 2006 and February 2009. Data were collected from all women in the 1st trimester of their pregnancy who had a live intrauterine pregnancy ( IUP) at the 1st transvaginal ultrasound scan ( TVS). 29 historical, clinical and ultrasound end points were recorded. Women were followed until the final diagnosis was established at the end of the 1st trimester: viability or nonviability. A multinomial logistic regression model was developed. The performance of this model was evaluated using receiver operating characteristic ( ROC) curves. Results Data from 416 pregnancies were included: 92.1% were live beyond the 1st trimester, and 7.9% had miscarried. The most useful prognostic variables for developing the logistic regression model were gestational age by dates, vaginal ( PV) bleeding, PV clots, gestational age by TVS, consistency with menstrual dates, mean gestational sac ( GS) size, mean yolk sac ( YS) size and number of previous caesarean sections. Used retrospectively on 416 women based on 25 imputations, the model gave an AUC of 0.88. Based on cross-validation, the independent predictive power obtained an AUC of 0.78. Conclusions We have developed a new model to predict the outcome of the 1st trimester in women with live IUP at the 1st scan. [ABSTRACT FROM AUTHOR]

This paper presents a two-stage procedure for estimating the conditional support curve of a random variable X, given the information of a random vector X. Quantile estimation is followed by an extremal analysis on the residuals for problems which can be written as regression models. The technique is applied to data from the National Bureau of Economic Research and US Census Bureau's Center for Economic Studies which contain all four-digit manufacturing industries. Simulation results show that in linear regression models the proposed estimation procedure is more efficient than the extreme linear regression quantile. [ABSTRACT FROM AUTHOR]

The aims of this study were to (1) compare prospectively the clinical and radiographic changes in periodontal and peri-implant conditions, (2) investigate the association of changes in periodontal parameters and peri-implant conditions over a mean observation period of 10 years (8–12 years) after implant installation, and (3) evaluate patient risk factors known to aggravate the periodontal conditions for their potential influence on the peri-implant tissue status. Eighty-nine partially edentulous patients with a mean age of 58.9 years (28–88 years) were examined at 1 and 10 years after implant placement. The patients contributed with 179 implants that were placed after comprehensive periodontal treatment and restored with crowns or fixed partial dentures. One hundred and seventy-nine matching control teeth were chosen as controls. Also, the remaining teeth ( n=1770) in the dentitions were evaluated. Data on smoking habits and general health aspects were collected at 1 and 10 years as well. At 10 years, statistically significant differences existed between implants and matching control teeth with regard to most of the clinical and radiographic parameters ( P<0.01) with the exception of plaque index (PII) and recession. Multiple regression analyses were performed to associate combinations of periodontal diagnostic parameters to the peri-implant conditions: probing attachment level (PAL) at implants at 10 years was associated with implant location, full-mouth probing pocket depth (PPD) and full-mouth PAL ( P=0.0001, r2=0.36). PPD at implants at 10 years correlated to implant location, full-mouth PPD and full-mouth PAL ( P<0.001, r2=0.47). Marginal bone level at implants at 10 years was significantly associated to smoking, general health condition, implant location, full-mouth PAL and change over time in full-mouth PPD ( P<0.001, r2=0.39). These results present evidence for the association between periodontal and peri-implant conditions and the changes in these tissues over 10 years in partially edentulous patients. To cite this article: Karoussis IK, Müller S, Salvi GE, Heitz-Mayfield LJA, Brägger U, Lang NP. Association between periodontal and peri-implant conditions: a 10-year prospective study. Clin. Oral Impl. Res. 15, 2004; – [ABSTRACT FROM AUTHOR]

Objective: To evaluate routinely applicable criteria to predict fragmentation of renal calculi by extracorporeal shock wave lithotripsy (ESWL). Patients and Methods: Two hundred and two consecutive patients (121 men, 81 women), median age 48 (range 19–81) years, were treated with the original Dornier HM-3 lithotriptor at a single stone center. Inclusion criteria were: solitary stones, 10–30 mm in greatest diameter, located in renal pelvis or calyces. Based on plain radiographs, the calculi were classified according to their size, form, location, density (compared to the 12th rib), structure and surface. Furthermore, age of the patient, gender and body mass index were also considered for evaluation. Disintegration was documented on day 1 after ESWL by plain X-ray. A multivariate regression analysis was applied to all preoperative parameters, based on the dual variable stone free versus residual fragments. Results: The overall disintegration rate was 95.5%; 42 patients (20.8%) were completely stone free, and 151 patients (74.7%) had clinically insignificant residual fragments (5 mm or smaller). 14.9% of men and 29.6% of women were stone free (p = 0.01). All other parameters did not reach statistical significance. Conclusions: The disintegration rate of the HM-3 is excellent for kidney stones; women did significantly better than men. However, because of this high disintegration rate, a much larger series would be necessary to define possible differences between preinterventional parameters, if there were any at all.Copyright © 2001 S. Karger AG, Basel [ABSTRACT FROM AUTHOR]