Logistic Joinpoint Regression Models

Abstract:  In this talk, a model which extends logistic regression to allow for one or more joinpoints will be considered. Maximum likelihood estimation of the model parameters will be discussed for the case when the number of joinpoints is known. Two multiple joinpoint algorithms will be described which can be used to estimate the number of joinpoints. Asymptotics of the maximum likelihood estimators and consistency of the associated confidence bounds produced by the parametric bootstrap are discussed for a clustered logistic joinpoint model with an unknown joinpoint. Finally, a couple of applications of these methods will be presented.