Statistics Seminar

Seminar: Dr. Frank Konietschke

Title: Permutation Tests for Unbalanced Heteroscedastic Factorial Designs
 
Dr. Frank Konietschke
Department of Medical Statistics, University Medical Center Göttingen
 
May 22nd
3:00 p.m.
MDS 335
 
Refreshments: 312 MDS building
 
Abstract: In many trials, data are collected in terms of a factorial design, e.g., when male and female patients are randomized to a>1 different treatment groups. Hereby, we are interested in testing the null hypotheses of no sex effect, no treatment effect, and / or no interaction between sex and treatment. Usually, the data are modeled by assuming homogeneous variances - an unrealistic assumption in higher way layouts. In particular, normality of the error term is often assumed. In several trials, however, the normality assumption is not justifiable, e.g. for skewed data.
In this talk, we consider inference methods (quadratic tests) for unbalanced heteroscedastic factorial designs under non-normality. Brunner et al. (1997) considered the so-called ANOVA-type statistic, which can be seen as an improvement over the classical Wald-type statistic. Both statistics, however, are quite poor in terms of controlling the type-I error rate under non-normality. Although the data are not exchangeable under heteroscedasticity, we investigate an unified permutation approach to achied valid procedures. We derive different conditional central limit theorems for the permuted statistics. Hereby, it will be shown that the conditional permutation dstributions are invariant under the main effects and interactions. The theoretical results verify the validity of the new approaches.  Extrensive simulation studies show that these permutation approaches greatly improve the standard procedures. A real data set illustrates the application.
Date:
-
Location:
MDS 335
Event Series:

Seminar: Dr. Matthew Gurk

 

Presented by Dr. Matthew Gurka, PhD of West Virginia University
Thursday, March 29th, 2012
 
Refreshments in MDS 206 3:30 – 4:00 PM
Colloquium in MDS 220 4:00 – 5:00 PM
 
ABSTRACT:
Child development and chronic illness are integrally related. The relationship between the health of the child and developmental outcomes has been an active area of research. Numerous studies have found associations between chronic illnesses, such as asthma, and behavioral problems in children. Analysis of data from a large longitudinal study, the National Institute of Child Health and Human Development (NICHD) Study of Early Child Care and Youth Development (SECCYD), allowed for further examination of these relationships throughout childhood. But, the required longitudinal data analysis motivated statistical research as well. Specifically, the initial study of childhood asthma and behavior illustrated how accurate inference for the fixed effects in the linear mixed model depends on the covariance model for the repeated measures. Simulation studies have revealed biased inference for the fixed effects with a mis-specified covariance structure, at least in small samples. We proved that incorrectly assuming a simple random intercept model (i.e., compound symmetry) leads to optimistically biased inference about the fixed effects, in both small and large samples. Simulations illustrate the bias and evaluate a variety of strategies aimed at avoiding it.
Date:
-
Location:
MDS 220
Event Series:

Seminar: Dr. Hao Wu

April 13th

4:00-5:00p.m.

MDS 220

Refreshments: 312 MDS building

Title:

Differential expression in RNA-seq

Abstract:

Recent developments in RNA-sequencing (RNA-seq) technology have led to a rapid increase in gene expression data in the form of counts. RNA-seq can be used for a variety of applications, however, identifying differential expression (DE) remains a key task in functional genomics. There have been a number of statistical methods for DE detection for RNA-seq data. One common feature of several leading methods is the negative binomial (gamma-Poisson mixture) model. The distinct feature in various methods is how the variance, or dispersion, in the gamma distribution is modeled and estimated. We evaluated several large public RNA-seq datasets and find that the estimated dispersion in existing methods does not adequately capture the heterogeneity of biological variance among samples. We present a new empirical Bayes shrinkage estimate of the dispersion parameters and demonstrate improved DE detection. 

 

Time permitting; I will present an on-going project to integrate protein binding ChIP-seq data in eQTL mapping. A hierarchical model is developed for the data integration. The prior probability for a SNP being associated with a gene can be modeled as a function of its surrounding protein binding profiles. Model parameters and posterior probabilities can be estimated via an EM type of algorithm.

Date:
-
Location:
MDS 220
Event Series:

Speaker: Dr. Kert Viele

 

Refreshments in POT 845 3:30 – 4:00 PM
 
Title:
Driving with your eyes open - A practical perspective on 
adaptive clinical trial
 
Abstract:
 
Over the past decade adaptive clinical trial designs have become an extremely "hot" area in the medical industry. In this talk we will review some of the basic principles of the adaptive design process both from a statistical perspective and from a logistical perspective (e.g. how does one interact with a clinical team to design and execute an adaptive clinical trial). We will illustrate techniques used to shorten trial duration, different criteria for stopping trials early, and mechanisms for adaptive allocation of subjects which allow more patients to be allocated to the most important arms of a study (arms which may not be known at the beginning of the study). We will also discuss the use of trial simulation in refining clinical trial design, which allows clinical teams to "test drive" the design to refine its operating characteristics and ensure the design answers the current research questions
Date:
-
Location:
CB 211
Event Series:

Speaker: Dr. Milan Studeny

http://staff.utia.cas.cz/studeny/studeny_home.html?q=user_data/studeny/studeny_home.html

Title: Algebraic approach to independence models and learning Bayesian networks

 

Abstract: 

The motivation for the talk is the description of probablistic conditional independence (CI) models

and learning graphical models. First, a quick overview of graphical approaches to the

description of CI structures will be given. Then the idea of algebraic description of CI structures will be

explained. It can be applied to computer testing CI implications using the methods of linear

programming. The rest of the talk will be devoted to a linear algebraic approach to learning Bayesian

networks, which are special graphical models. The core will be a report on recent results related

to the aim to apply the methods of integer programming in this area.

 

 

 

Date:
-
Location:
Classroom Building Rm. 102
Event Series:

Seminar: Dan Weiner

Pharmacometrics as an Emerging Discipline – Application and Utilization to

Improve Decision Making in the Pharmaceutical Industry

Daniel L. Weiner, PhD

SrVP and CTO, Certara Corporation

Adjunct Associate Professor, Division of Pharmacotherapy and Experimental Therapeutics,

School of Pharmacy, College of Medicine, UNC Chapel Hill

Nonlinear mixed effects modeling (NME) has been a mainstream application in the pharmaceutical

industry for over 20 years. Initially NME was applied after an NDA submission was submitted, and was

used to support dosing regimens in special populations (e.g., elderly, pediatric). Despite the advance of

new technologies for identifying molecules for further development, and methods to assess their utility

(e.g. biomarkers) the failure rates of drugs late in clinical development remain persistently high. As a

result, companies are attempting to derive a deeper understanding of the mechanism of action of drugs

and the underlying diseases they are trying to treat. These newer methods, collectively denoted as

Pharmacometrics, utilize deeper biological modeling than previously applied using NME, along with

disease progression modeling, and extensive simulations based on in-silico models of a drug’s

pharmacokinetics and pharmacodynamics; drug development plans; and the individual protocols to be

conducted.

During this talk these topics will be discussed, along with a few case studies illustrating the application of

Pharmacometrics to drug development.

Date:
-
Location:
TBD
Event Series:

Yuguo Chen (from U of Illinois at Urbana-Champaign)

Sampling for Conditional Inference on Network Data

Random graphs with given vertex degrees have been widely used as a model for many real-world complex networks. We describe a sequential sampling method for sampling networks with a given degree sequence. These samples can be used to approximate closely the null distributions of a number of test statistics involved in such networks, and provide an accurate estimate of the total number of networks with given vertex degrees. We apply our method to a range of examples to demonstrate its efficiency in real problems.
 
Personal webpage:
Date:
-
Location:
CB 102
Event Series:
Subscribe to Statistics Seminar