 |
|
|||||
|
|
The 2007 Annual Case Western Reserve University Cleveland Clinic Foundation Ohio State University Joint Biostatistics Symposium
Thursday May 17, 2007 Case Western Reserve University Parking and Walking on Campus Directions ____________________________________________________ AGENDA School of Medicine Room T501 11:15 – 11:45 Registration and Parking Validation
11:45 – 12:45 Lunch
School of Medicine Room E501 1:00 – 1:45 Bo Lu, Ph.D. with Ohio State University Exploring Propensity Score Matching in Longitudinal Studies
1:45 – 2:30 Denise Babineau, Ph.D. with the Cleveland Clinic Foundation Goodness of Fit Tests for Parametric Models Applied to Interval-Censored Survival Data Interval censored data arise when subjects are observed intermittently such that their failure time is observed to lie only between two successive inspection times. In an attempt to model the failure time distribution, parametric models may be fit. In this situation, a fundamental question that must be addressed is the appropriateness of the fitted parametric model to the underlying failure time distribution. A Pearson goodness of fit test is a common approach to this problem whereby observed and expected failures in non-overlapping time intervals are compared. Due to overlapping time intervals in interval censored data, such a test is not directly applicable and methods that either define pseudo observed or pseudo expected failures in a set of time intervals must be used. This talk will discuss various methods for defining pseudo observed or expected failures, with particular focus on defining a pseudo expected failure using a nonparametric estimate of the observed inspection process.
2:30 – 2:45 Break
2:45 – 3:30 Jiayang Sun, Ph.D. with Case Western Reserve University Feature Selection: Overview and Current Research Variable or feature selection is an important step in regression, discriminant and cluster analyses (a.k.a. supervised and unsupervised learning in machine learning), especially when the number of features is large, or the nuisance features are influential. In this talk, I will discuss variable and feature selection procedures via a unified framework, review challenges for large and online data, and present our four current research projects on feature selection in supervised and unsupervised learning. The ideas of ranking features, or sets of features will be presented. Coherence indices for clusters will be introduced. Subsampling for principal component analysis, adaptive mixture for feature selection will be studied. Applications to some microarray data will be presented. Part of the talk is based on the joint work with Peng Liu, Yaomin Xu and Zhongfa Zhang.
3:30 – 4:30 Keynote speaker: Cyrus Mehta, PhD. with Cytel, Inc. Flexible Adaptive Trial Design: Two Case Studies An adaptive trial is one in which interim data from the trial itself is used to modify and improve the study design, without undermining its validity or integrity. Trial sponsors and regulators have expressed a great deal of interest in designing such trials because of their potential benefit for Phase II and Phase III programs. In the Phase II setting an adaptive trial can assign a larger proportion of the enrolled subjects to the treatment arms that are performing well, drop arms that are performing poorly, and investigate a wider range of doses so as to better identify the nature of the dose-response relationship and select doses that are most likely to succeed at Phase III. When the trial proceeds to Phase III an adaptive design can facilitate early identification of efficacious treatments, determine if the trial could be terminated for futility, and make adjustments to sample size or follow-up duration so as to ensure that the trial is adequately powered. In some cases it might even be possible to enrich the patient population by altering the eligibility criteria at an interim look. Thus, adaptive trials have the potential to translate into more ethical treatment of patients within trials, more efficient drug development, and better focusing of available resources. On the other hand, such trials require tremendous up-front planning and simulation to verify their operating characteristics, precisely because they are so flexible. In this seminar we give an overview of adaptive clinical trials, pointing out their advantages as well as their limitations. The presentation will be conceptual rather than technical and will be illustrated with two case studies of actual trials drawn from our own consulting experience. Logistical and regulatory issues will be discussed.
4:30 Adjourn _________________________________________________________________________________________________________________________________________________ Co-Sponsors Cytel, Inc. Glaxo-Smith-Kline Merck Research Laboratories Department of Epidemiology and Biostatistics (CWRU)
Registration deadline has been extended to May 4th, 2007. RSVP to donna.marine@case.edu
|
|||||
