A Review of Classification Methods That Could Be Used to Identify a Subset of Patients in a Clinical Study

• Activity Number: 356
• Type: Contributed
• Date/Time: Tuesday, July 31, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Alexander Cambon*+ and Shesh N. Rai
• Companies: University of Louisville and University of Louisville
• Keywords: classification ; machine learning ; dimension reduction

Abstract

It is now widely recognized that many treatments for cancer are effective for a subset of a population. However large clinical studies for cancer treatments are powered to detect an overall treatment effect. In this study, classification methods which could be used in a clinical study setting to identify subsets of patients which are more sensitive to treatment or more prone to relapse are reviewed. Some approaches to sample size planning for classifiers, quantification of uncertainty, and methods which assess added value of genomic/high dimensional data to clinical covariates are also highlighted. Classification methods such as logistic classification, Random Forests, and Boosting will be studied.