I am primarily interested in methodology for and applications of survival analysis. My thesis deals with inference on competing risks data, where there is more than one failure type. Competing risks occur frequently in medical applications. For example, in breast cancer clinical trials, a number of different events may occur, which may be classified into local-regional events and all other events.
My work has developed parametric methods for competing risks data. As cause-specific cumulative incidence functions (similar to the cumulative distribution function, though specific to a particular event type) are by their nature improper, I have sought to develop distributions which are more appropriate for competing risks data. Additionally, current methods of inference make specific assumptions about the nature of the relationship between the failure times of each of types of events. I have also developed a joint modeling method which allows investigators to make inference about the relationship between failure times for different event types.
My thesis was successfully defended on July 22, 2008. A copy of my dissertation is available as a pdf document.