Department of Statistics
University of California-Irvine
“Statistical Modeling for Biomedical Applications”
In this talk, I will discuss two recent projects motivated by biomedical applications. The first project is about classification of image data using low-rank regularization. Theoretical properties such as risk bound and rank consistency are studied. The second project is about risk prediction for liver cancer. I will discuss estimation of time-dependent characteristic curves and construction of a score system that combines the information of biomarkers with other baseline covariates. I will also discuss an ongoing work on hypothesis testing for the number of components in the mixture model.