Longitudinal Modeling when the Response and Time-Dependent Covariate(s) are Measured at Distinct Time Points

Joel A. Dubin
Department of Statistics & Actuarial Science
Department of Health Studies & Gerontology
University of Waterloo

Thursday, September 10, 2009
3:30pm
Mayo 3-100
Minneapolis Campus

Abstract:

In this talk, we will discuss a flexible method to handle both association and temporal sequencing of distinct longitudinal measures, where the measures may be of mixed type (e.g., one continuous, the other binary) and recorded on non-uniform grids and different time points from one another. A smoothing step will be involved, and, in particular, a binning approach is proposed. This approach will be demonstrated on a dataset of hemodialysis patients, where longitudinal measures of health outcomes (e.g., infection) were recorded at different time points than longitudinal physiologic measures such as serum C-reactive protein levels (a marker for inflammation). An interesting scientific question to answer is whether experiences of infection follow or predate inflammation, or if any association detected can only be seen contemporaneously. We will also provide simulation results suggesting when binning may be recommended for this type of data.

A social tea will be held at 3:00 P.M. in A434 Mayo. All are Welcome.
For more details contact 612-624-4655 or see http://www.biostat.umn.edu/seminar_academic.html