The University of Texas MD Anderson Cancer Center
Department of Biostatistics

My research career has been characterized by a progression from a broad theoretical and methodological program at Purdue University with a focus on engineering applications, experimental design, and experimental biology, to a health care focus at University of Texas Medical Branch (UTMB), where I have focused on clinical trials in aging and pediatric burns, and the development of otitis media in infants. 

Recently, my work has become focused on improving the performance of machine learning applications for which overfitting appears in unexpected ways, sample size estimation, and joint modeling of longitudinal factors and correlated event outcomes such as death or drop-out.  Each of these investigations stems from individual application projects with complex data structures.

One example stems from a project wherein I developed procedures to overcome the bias of power calculations and sample size determinations based on pilot data or interim data.  These biases are not recognized for problems in which a target power is desired.

I also have presented modified reliability indexes, with improved performance and modeling structure.  The underlying model can be generalized to a variety of uses.  In particular, this modeling procedure improves the estimation of learning models with potential category misidentification.

PubMed

Education & Training

PhD, Stanford University, 2003