A semi-Markov model for multi-state and interval-censored data: An application
of the cognitive states present in the Uniform Data Set
 
   Lead Investigator:    Jeremiah Aakre
   Institution      :    Mayo Clinic ADRC
   E-Mail           :    aakre.jeremiah@mayo.edu
   Proposal ID      :    398



   Proposal Description:
   Semi-Markov models are often considered for disease progression data when the 'memory-less' Markov assumption may be considered too strong.  In semi-Markov analyses, emphasis is placed on time spent in a state, or the waiting time before a transition into a different state.  In addition, cognitive states are not always considered to be exactly defined or right censored.  Instead the transition times are known to have occurred in a given interval censored.  We will consider semi-Markov models which  allow for interval censoring by a parametric framework.  Covariates will be considered for each transition separately to allow for sparse models.  The overall goal will be to explore specific factors and their associations with transitions between cognitive states.