Spring 2006 Meeting

Held on March 9, 2006 at the Wyndham Glenview Suites.

The Program consisted of four presentations:

  1. Modern Statistical Analysis Using Stata Roberto Gutierrez, Ph.D., StataCorp LP  
  2. Statistical and Graphical Analysis of Clinical and Observational Data Using Splus Michael O’Connell, Ph.D. Insightful Corporation  
  3. Statistical Analysis Using SPSS  Ming-Long Lam, Ph.D. SPSS Inc.  
  4. SAS Enterprise Miner, SAS Forecast Server, and Other SAS Statistically Based Solutions William T. Winand and Ross Bettinger 


Modern Statistical Analysis Using Stata by Roberto Gutierrez, Ph.D., StataCorp LP

Biographical Background Dr. Roberto Gutierrez received his PhD in Statistics in 1995 from Texas A&M University. He is StataCorp's Director of Statistics. His area of specialization is in survival analysis and mixed models.

Abstract Stata is a widely-used general purpose statistical software package for data management, statistical analysis, and graphics. Currently at version 9.1, Stata is particularly suited for biostatisticians and epidemiologists. Besides the standard statistical routines such as linear regression, ANOVA, and regression models for limited dependent variables, included in Stata are tools for fitting linear mixed models, analyzing epidemiological tables, and survival analysis.  Stata also possesses an easy-to-use GUI interface, the ability to produce publication-quality graphics, and is based upon a fully-integrated programming environment that includes Mata, a matrix programming language. This discussion will look at some of the Stata capabilities for modern statistical analysis.


Statistical and Graphical Analysis of Clinical and Observational Data Using S-plus by Michael O’Connell, Ph.D. Insightful Corporation

Biographical Background Dr. Michael O'Connell has been working in the medical device, diagnostics, pharmaceutical, and biotech arena for the past 15 years. He has recently concentrated his efforts in two areas: statistical analysis of safety data and statistical analysis of microarray data. In both of these areas, O’Connell is involved in the design and development of tools for analysis and reporting of clinical and discovery data from S-PLUS. Dr. O'Connell, Director of Life Science Solutions for Insightful Corporation, has been also named one of the “100 Most Inspiring People in the Life Sciences Industry” by the readers of PharmaVOICE magazine. Dr. O'Connell holds a Bachelors degree in Science from the University of Sydney, a Masters degree in Statistics from the University of New South Wales and a Ph.D. in Statistics from North Carolina State University.

Abstract Prospective, randomized, double-blind clinical trials enable inference on causality. Retrospective observational studies, e.g. using medical claims data, have a number of limitations compared with randomized prospective clinical trials, most notably with respect to balance in known and unknown prognostic factors which may cause a preference for one treatment over another. This talk provides examples of statistical and graphical analysis of clinical trial and observational data. The clinical examples include interactive graphical review of results with end-users, graphics in formal clinical study reports, and comparisons of adverse event data using Bayesian hierarchical models. The observational examples include comparison of length of stay for patients receiving new and standard therapy for treatment of a variety of infection conditions using medical claims data from over 500 hospitals. This analysis uses logistic regression and propensity scoring; and includes 2-sample and survival curve comparisons, combined with graphical analysis before and after the matching, assessing covariate balance and treatment effects on length of stay.

The examples feature the use of S-PLUS Trellis graphics using multiple sources of data and metadata; and the S+Graphlet® technology with S-PLUS Server for interactive graphical analysis and data browsing. S-PLUS provides an extensive set of tools for statistical and graphical analysis of clinical and observational data.


Statistical Analysis Using SPSS by Ming-Long Lam, Ph.D. SPSS Inc.

Biographical Background Dr. Ming-Long Lam is currently the manager of Statistics Research and Master Statistician at SPSS Inc. in Chicago. Ever since he joined SPSS in 1992, he led development of many statistical procedures for analyzing the General Linear Models, the Variance Component Models, the Mixed Effects Models, the Multinomial Logistic Models, the Ordinal Regression Models, the Two-Step Cluster Model, the Complex Samples module, and the forthcoming Generalized Linear Model. Besides his work at SPSS Inc, he and a co-author will publish the book "Using Data Analysis to Improve Student Learning: Toward 100% Proficiency". This book will be available in September 2006. He received his Ph.D. degree in Statistics from the University of Chicago.

Abstract Recent versions of SPSS were released with a significant improvement in features that enhance the process of performing statistical analyses, enrich the presentation of results, and provide new or improved analytical procedures. This discussion will highlight features that may impact the work of statisticians in enhancing various statistical analyses, both simple and complex. Some examples and applications will be drawn from mixed models, multinomial logistic models, and programmability feature. A sneak preview of the forthcoming generalized linear model will be given too.


SAS Enterprise Miner, SAS Forecast Server, and Other SAS Statistically Based Solutions by William T. Winand and Ross Bettinger 

Biographical Background Mr. William T. Winand is a Senior Systems Engineer at SAS. William is Spearhead for Analytical Intelligence within SAS' Health & Life Sciences Organization. In this role, William consults with pharmaceutical, biotechnology, medical device, and health insurance companies to identify areas where analytics can provide significant business value and to demonstrate the capabilities of SAS Enterprise Miner, SAS Forecast Server and other SAS statistically based solutions to deliver that value. William has worked for SAS since 1995. William also holds a Masters of Management degree from the Kellogg Graduate School of Management, Northwestern University. 

Mr. Ross Bettinger is an Analytical Consultant at SAS Institute. He obtained his B.A. degree in Mathematics from UCLA, and Master’s degrees in Systems Engineering from UCLA, Business/Statistics from University of Wisconsin, Madison, and Electrical Engineering from Northeastern University. He has over twenty-plus years of SAS® experience and eleven years of statistical modeling experience. His area of expertise is in statistical analysis, forecasting, data mining, and text mining. Mr. Bettinger has been involved in various consulting projects including credit card risk management, modeling in direct marketing, financial analysis and forecasting, text mining for customer retention modeling in banking, and warranty analysis. Some specific research questions that Mr. Bettinger has been involved include work in credit card risk management settings to identify key events that signal a card member’s likelihood to default on the card’s remaining balance and bad debt, developing neural network models for cross-sell and up-sell modeling for a credit card issuer, writing specialized software that substantially reduced the scoring time required for very large volumes of records, modeling card members’ spending patterns and propensity to run up a high balance, building time series models of future levels of bad debt, and analyzing call center data to determine if there were specific words associated with accountholder attrition. Prior to joining SAS, Mr. Bettinger worked as an Analytical Consultant for Epsilon Data Management, Sears, Roebuck & Co., and Magnify, Inc. where he built models for up-sell and cross-sell campaigns, catalog sales, insurance underwriting, and credit card risk management.

Abstract Data mining is the process of data selection, exploration and building models using vast data stores to uncover previously unknown patterns. What does this mean to you, and what is its value your organization? With organizational data growing exponentially, data mining is now a necessary tool to shorten time to discovery and to provide new insights. SAS Enterprise Miner, the industry leading data mining solution, streamlines the data mining process to create highly accurate predictive and descriptive models based on analysis of vast amounts of data from across an enterprise. Examples from different areas of the Health and Life Sciences organizations will be used to demonstrate the capabilities and value of SAS Enterprise Miner.