Department of Biostatistics
John Hopkins University
Univariate and Bivariate Recurrence Time Data: Tricky Data and a Few
Solutions
Recurrent event data are frequently encountered in longitudinal follow-up
studies. In statistical literature, methods and inferences for recurrence
time data were developed mainly for parametric models. Standard
nonparametric and semiparametric techniques (Kaplan-Meier estimate, weighted
log-rank tests, partial likelihood, etc.) have been used in many medical
papers to analyze recurrent event data but the adopted approaches are often
inappropriate. These inappropriate approaches could result in analyses with
small or significant bias, depending on the data structure. Suppose the
outcome measures of interest are the (bivariate) recurrence time(s) between
successive events. This talk will consider nonparametric estimation of
survival functions from univariate and bivariate recurrence time data. Some
probability characteristics of recurrence times will be discussed for the
understanding of how to construct the proposed estimation procedures. The
performance of the estimators will be examined by simulations and an
analysis from schizophrenia data.
About the Speaker: Dr. Mei-Cheng Wang is currently Professor in the
Department of Biostatistics of Johns Hopkins University. She obtained her
Ph. D. from the University of California at Berkeley. Recently she has been
elected a Fellow of the American Statistical Association. Her research areas
are in the topics of nonparametric and semiparametric analysis for multiple
event data, inferences for truncated or prevalent cohort data,
semiparametric conditional inferences, statistical methods for multivariate
case-cohort studies, and competing risks models for epidemiological studies,
and has numerous publications in these areas.
Back to Colloquium Series