CANCELLED: A semi-Markov approach to modelling age-dependent survival

[This talk has been cancelled -- it will be rescheduled for semester 2]

For many species, survival varies with age.  For example, first-years may have a particularly high mortality rate compared to the population as a whole, or there may be different age-classes such as young, adult and senior.  Additionally, survival, capture and recovery probabilities may also be time-dependent, hence fully age-dependent models can lead to a significant increase in the number of model parameters.  The specification of any age-dependence on the survival probabilities induces a distribution on the age-structure of the population and hence also the distribution of the age of death of individuals.  We propose a semi-Markov survival process by specifying a parametric model for the age of death.  This approach allows for the incorporation of a sensible age-structure for the species under study and can be fitted using efficient hidden Markov model methods.