99
reads

Ching-Hsi Lee
2010-04-09  12:30 - 14:30
Room 405, Mathematics Research Center Building (ori. New Math. Bldg.)

With the great advance in the discovery of biomarkers, recent research interests focus on seeking approaches for combining biomarkers. In many practices, it has been found that a single biomarker is not sufficient to serve as a device for the early stages of prediction. In this discussion, I’ll introduce their model-combining algorithm, which mainly utilizes the idea of adaptive regression by mixing with screening (ARMS), for classification in biomarker studies. It works by weighted over various combinations of linear predictors under different weighting schemes. The proposed weights in the algorithm are further justified through decision theory and risk-bound results.