Seminars

Thursday, January 17, 2008
4:30 - 5:30 p.m.
UA Whitaker Bldg, School of BME, Room 2110
SPEAKER: Dr. Melinda Higgins, GTRI and Emory University
TITLE: A Practical Approach to Power, Effect Size and Sample Size & An Overview of PASS (Power Analysis and Sample Size) Software
ABSTRACT: This lecture will cover: an overview of Statistical Power (theory and assumptions) in easy to understand language; an introduction and overview of the PASS software (menus, help manuals and output structure); the various parameters affecting power such as "effect size," sample size, and others; several detailed examples including examples using proportions, means and correlations as well as highlight all of the various statistical models for which power can be assessed currently in PASS; and a few other statistical power software packages available (including some free ones).

SLIDES in PDF.

 

Thursday, February 7, 2008
4:30 - 5:30 p.m.
UA Whitaker Bldg, School of BME, Room 3115
SPEAKER: Dr. Melinda Higgins, GTRI and Emory University
TITLE: Latent Variables (Variates) and Multicollinearity: Good or Bad? ABSTRACT: The majority of statistical modeling methods begin with an assumption that the independent variables (predictors) be independent. However, true independence is almost never achieved. Some correlation (or co-linearity) is usually present. Therefore, it is important to understand how to measure the extent of non-independence as well as knowing how to handle/correct for highly correlated/related variables.
This lecture will cover this material with minimal equations and minimal math theory and will instead explain the concepts through graphical techniques and visualizations. The following topics will be addressed:

• Multicollinearity - what it is as well as disadvantages and advantages of multicollinearity;
• Latent Variates - eigenvalues and eigenvectors explained;
• IRIS Dataset - this classic dataset will be described and analyzed through multi-dimensional visualization tools and real-time animations;

This lecture will also compare and contrast the following analysis methods with summaries on when to use which method as well as advantages and disadvantages of each method:

• Principal Components Analysis
• Factor Analysis - VARIMAX rotation example - discussion on extracting "underlying constructs;" and
• Discriminant Analysis

SLIDES in PDF

Thursday, February 14, 2008
4:00 - 5:00 p.m.
MS&E Building, Room 1224
SPEAKER: Dr. Tianwei Yu, Biostatistics at Emory University
TITLE: LC-MS Data Pre-processing for Metabolomics --- An Adaptive Strategy
ABSTRACT: Liquid chromatography-mass spectrometry (LC-MS) profiling is a promising approach for the quantification of metabolites from complex biological samples. Significant challenges exist for the analysis of LC-MS data, including noise reduction, feature identification and quantification, and data alignment. Vast amounts of data requires the algorithms to be highly efficient. Different machines and/or running conditions may also yield data that differ in characteristics. Here we present a package written in R for the processing of such data. The major technical improvement over existing methods is the adaptive tolerance level searching rather than hard-cutoff or binning. In addition, the EM algorithm is used to better estimate peak intensities when m/z sharing occurs. We also use a run-filter to replace the signal/noise level cutoff to better preserve weaker signals. The algorithm written in R achieves reasonable speed processing large LC/ MS datasets.

Tianwei Yu received his Ph.D. degree in statistics from UCLA in 2005. Before his PhD training, he received undergraduate and graduate (M.S.) training in Biology. He is currently an assistant professor at Emory Department of Biostatistics. His research is focused on Bioinformatics, Statistical Genomics, Metabolomics and biological network modeling.

Thursday, February 21, 2008
4:00 - 5:00 p.m.
MS&E Building, Room 1224
SPEAKER: Dr. Andre Rogatko Winship Cancer Institute at Emory University
TITLE: Evaluation of Patient Characteristics as Predictors of Acute Treatment Toxicity
ABSTRACT: We will discuss EWOC (Escalation with Over Dose Control), the first statistical method to directly incorporate formal safety constraints into the design of cancer phase I trials. The method controls the frequency of overdosing by selecting dose levels for use in the trial so that the predicted proportion of patients administered a dose exceeding the MTD (Maximum Tolerated Dose) is equal to a specified upper bound. We will also discuss an extension of EWOC that permits the utilization of information concerning individual patient differences in susceptibility to treatment. This is the first method described to design cancer clinical trials that not only guides dose escalation but also permits personalization of the dose level for each specific patient. The method adjusts doses according to patient-specific characteristics and allows the dose to be escalated as quickly as possible while safeguarding against overdosing. The extension of EWOC to covariate utilization was implemented in five FDA approved phase I studies that will be also discussed.

Andre Rogatko [ http://sisyphus.emory.edu/andre.html ] is a Professor of Biostatistics at Emory's SPH and Director for Biostatistics Research and Informatics at Winship Cancer Institute. Dr Rogatko is one of the world leaders in Bayesian clinical trials. He graduated with BS, MS and PhD at University of Sao Paulo, Brasil and after graduation was a affiliated with Cornell University, Sloan-Kettering Cancer Center, World Health Organization, Fox Chase Cancer Center, prior to joining Emory University and Winship Cancer Institute. Dr Rogatko has an extensive publishing record and outstanding service to the field and profession.

Friday, March 7, 2008
1:00 - 2:00 p.m.
TBA
SPEAKER: Dr Pulak Ghosh, Georgia State University
TITLE: Joint Modeling of Longitudinal Data and Informative Dropout in the Presence of Multiple Changepoints with an Application to HIV-AIDS
ABSTRACT: In longitudinal studies of patients with the Human Immunodeficiency Virus (HIV), objectives of interest often include modeling of individual-level trajectories of HIV Ribonucleic Acid (RNA) as a function of time. Such models can be used to predict the effects of different treatment regimens or classify subjects with similar trajectories into subgroups. This, in turn, helps in determining the optimal treatment combination(s), which is an important part of drug development. Empirical evidence, however, suggests that individual trajectories often possess multiple points of rapid change, which may vary from subject to subject --- both in number and in location. Presence of such changepoints the modeling of individual viral RNA levels becomes difficult, since usual methods become unsuitable. In this talk, we develop a new robust multiple-changepoint model which satisfactorily addresses the above issues. The proposed method uses a joint model to incorporate information from the longitudinal data as well as from informative dropouts, which are common in such studies. A Dirichlet process prior is used to model the distribution of the changepoints of individual trajectories. The inherent clustering property of Dirichlet process leads to a natural clustering, and thus, sharing of information among subjects with similar trajectories. A fully Bayesian approach is used for model fitting and prediction and is implemented using the Gibbs sampler. The proposed method is illustrated using data from the ACTG 398 clinical trial. The results suggest that the proposed method is a significant improvement over the currently available methods.

After obtaining his PhD in statistics from Oakland University in 2003, Dr Pulak Ghosh joined Georgia State University, Department of Mathematics and Statistics. His research interests cover biostatistics, Bayesian statistics, modeling social network and behavioral data, clinical trials, and several other areas. Dr Ghosh publishes in top statistical and applied journals, has successful funding record and an active involvement in the academic community (professional societies, editorship and refereeing, grant evaluation, etc.).