Generalized longitudinal data analysis with applications to evaluating hospital utilization based on administrative databases.
There are many practical situations where subjects can experience recurrence of an event, the event has certain duration, and both the rate of event occurrences and the cumulative event duration are of particular interest.
Well-developed methods for recurrent events analysis do not take into account the event duration, which could lead to undesirable inferences. Motivated partly by the research project with the BC Cancer Agency to evaluate the hospital utilization of young cancer survivors, we develop a method to analyze recurrent event data with adjustment for event duration. Our methodology can be viewed as an extension of the well-established approaches for recurrent events. We also develop an approach to fitting semi-parametric models for a general response process, which includes counting processes as a special case. Data from the cancer project are used through the thesis to illustrate our formulation and approaches.
This type of interdisciplinary work is a hallmark of our program in Applied Statistics at Simon Fraser University. For more information, please contact Celes Ying (firstname.lastname@example.org) or her supervisor Joan Hu (email@example.com), Department of Statistics and Actuarial Science.
28 July 2006