Cleveland Clinic Foundation
Bunts Auditorium, Jennings Education Bldg, Area TT
E. 90th St. between Euclid Ave and Carnegie Ave.

Schedule:

Noon - 12:15pm Registration in Bunts Auditorium Lobby
12:15 - 1:15pm Lunch in Bunts Auditorium Lobby
1:15 - 2:00pm - Liang Li, Cleveland Clinic
2:00 - 2:45pm - Yoonkyung Lee, Ohio State University
2:45 - 3:00pm - Break
3:00 - 3:45pm - Jeffrey Albert, Case Western Reserve University
3:45 - 4:45pm - Guest Speaker, Donald Rubin, Harvard University

Directions and Parking:
At Cleveland Clinic Parking Lot #1, off E. 93rd St. between Euclid Ave and Chester Ave. See CCF web page for driving directions and campus map showing parking and the Symposium location (Area TT).

Titles and abstracts:

Liang Li - CCF
Regression Model with Heteroscedastic Covariate Measurement Error
Abstract:
Glomerular filtration rate (GFR) is an important biomarker in studies of chronic renal disease. However, it is difficult to measure, and the measurements may be subject to considerable measurement errors. As part of our efforts to develope and validate accurate estimating equations for GFR, we studied the measurement error and its effect on linear regression. Unlike most other measurement error models where the error variance is a constant, the error variance of GFR seems to depend on the actual level of GFR, and the variance can not be adequately stabilized by monotone transformations. We propose a class of heteroscedastic error models for GFR, and fit these models with a unified approach based on unbiased estimating equations. We will illustrate our model and method with examples that we are currently working on.

Yoonkyung Lee - OSU
A Sparse Solution Approach to Gene Selection for Cancer Diagnosis Using Microarray Data
Abstract:
Cancer diagnosis or prognosis based on gene expression profiles has been studied as a potentially more accurate means of predicting the disease status than standard methods based on histological observations. Presence of much larger number of genes than the sample size in the problem poses a challenge in building reliable and interpretable classification schemes. This talk will present a sparse solution approach for simultaneous gene selection and classification via component penalization of Support Vector Machines. The proposed method selects relevant genes in a principled way by taking into account their joint effects, remedying the limitation of common approaches of filtering genes marginally. Real data analysis will be given for illustration of the method and related issues will be discussed.

Jeffrey Albert - CWRU
Causal Models for Assessing Mediating Variables in Longitudinal Studies
Abstract:
In longitudinal clinical research there is often interest in the causal pathway through which treatment or exposure affects the response variable. When intermediate variables are available, a corresponding question is whether, and to what degree, certain intermediate variables mediate this treatment effect. Despite frequent interest in this question, there are few and limited statistical methods available for assessing mediation. To provide a foundation for causal inference, we propose a definition of mediation in terms of potential outcomes. Using this definition, we present measures of mediation in terms of parameters from a linear structural or structural equations model. Using two-stage least squares estimation of model parameters, we derive inference methods for mediation. The robustness to model assumptions is discussed and generalized methods considered. The new methods are applied to a study of the effect of music on post-surgical pain.

Donald Rubin - Harvard
Causal Inference Through Potential Outcomes: Application to Quality of Life Studies with "Censoring" Due to Death and to Studies of the Effect of Job-training Programs on Wages
Abstract:
Causal inference is best understood using potential outcomes, which include all post treatment quantities. The use of potential outcomes to define causal effects is particularly important in more complex settings, i.e., observational studies or randomized experiments with complications such as noncompliance. Here we deal with the issue of estimating the casual effect of a treatment on a primary outcome that is "censored" by an intermediate outcome, for example, the effect of a drug treatment on Quality of Life (QOL) in a randomized experiment where some of the patients die before their QOL can be assessed. Because both QOL and death are post-randomization quantities, they both should be considered potential outcomes, and the effect of treatment versus control on QOL is only well-defined for the subset of patients who would live under either treatment or control. Another application is to an educational program designed to increase final test scores, which are not defined for those who drop out of school before taking the test. A further application is to studies of the effect of job-training programs on wages, where wages are only defined for those who are employed, and thus the effect of the job-training program on wages is only well-defined for the subset of individuals who would be employed whether or not they were trained. Some empirical results are presented from Zhang, Rubin and Mealli (2004), which indicate that this framework can lead to new insights because the analysis is not predicated on traditional econometric assumptions.