Algorithms for Prediction and Subgroup Formation in Personalized Medical Decision-Making

• Activity Number: 595
• Type: Topic Contributed
• Date/Time: Wednesday, August 12, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Kelci Miclaus* and Richard Zink and Russ C. Wolfinger
• Companies: SAS Institute and SAS Institute and SAS Institute
• Keywords: predictive modeling ; personalized medicine ; subgroup analysis ; model selection ; cross-validation

Abstract

With recent technology advances, it has become standard practice to collect more and more data, especially 'omics-based biomarkers, on subjects enrolled in clinical trials. This trend has led to an expansion of methods to utilize biomarkers during clinical development, and tailor treatment decisions in medical practice. Predictive models are being applied to create better diagnostic tests, determine optimal treatment assignments based on safety and efficacy responses, and perform trial adaptations. In this presentation we will discuss a framework to navigate through millions of possible predictive model fitting procedures and make method comparisons via cross-validation to avoid overfitting. We also examine the application of predictive models for subgroup analysis in order to identify individuals that may receive enhanced benefit or increased harm by varying treatment options.