Methods for Assessing Pulsatile Associations Between Two Hormones

• Activity Number: 17
• Type: Contributed
• Date/Time: Sunday, August 11, 2002 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section*
• Author(s): Nichole Carlson*+ and Timothy Johnson and Morton Brown
• Affiliation(s): University of Michigan and University of Michigan and University of Michigan
• Keywords: hormone associations ; Bayesian approach ; model comparison

Abstract

The relationships between hormones regulate many biological systems. For instance, when hormones are secreted in pulses, a pulsatile release of one hormone may be associated with a pulsatile release of another hormone. We term this a driver-response relationship. To understand pulsatile relationships, blood samples are collected for an extended period of time and assayed for various hormone concentrations. The result is a time series of hormone concentrations for each subject and each hormone assayed. We have developed a model of a one-to-one, pulsatile driver-response association between two hormones. We write this model in the Bayesian framework and use Stephens' (2000) approach, an alternative to reversible jump methods, to jointly estimate the number of pulses and the distributions of other relevant parameters. This talk is focused on the results from a simulation study assessing the performance of this model under two model misspecifications:1.) non one-to-one pulsatile associations and 2.) no pulsatile association. The conclusions will be used to develop methods for testing the one-to-one pulsatile association assumption.