UW Madison UW School of Medicine and Public Health
UW Biiostatistics & Medical Informatics UW Biostatistics & Medical Informatics UW Madison UW Biostatistics & Medical Informatics Site Map


General Departmental Seminar Series


Probabilistic Modeling and Non-and Semi-Parametric Inference for
Event Times in Recurrent Event Settings

Professor Edsel A. Pena
Department of Statistics
University of South Carolina, Columbia

Wednesday, November 7, 2007
4:00 pm
140 Bardeen



In a variety of situations in the engineering, biomedical, public health, economic, 
political, and sociological settings, the event of main interest is recurrent. Examples
are failure of a subsystem in a machine (e.g., the space shuttle), occurrence of a tumor,
occurrence of migraine, the Dow Jones Industrial Average (DJIA) decreasing by at least
200 points on a trading day, and serious disagreements between a married couple.
Of statistical interest in such studies is to determine or estimate the inter-event distribution,
possibly in relation to observed covariates. Knowledge of this inter-event distribution,
or its relationship with covariates, could be helpful in lessening the impact of deleterious events,
or possibly increase the gain from beneficial events. In this talk, I will first provide
a brief survey of the nonparametric estimation of the event time distribution for single-events,
then proceed with the non-parametric and semi-parametric estimation of the inter-event time
distribution in the recurrent event setting. This will entail describing recently proposed dynamic
stochastic models for recurrent events which take into account different aspects of recurrent
event accrual such as the impact of interventions and covariates, as well as the impact
of the so-called sum-quota accrual scheme, which arise due to a finite monitoring window.
In the process, some discussions will also be provided of the advantages and disadvantages
of full versus marginal modeling approaches to multiple event settings.

Return to seminar list