Bivariate Mixture Cure Rate Model with Moran-Downton Weibull Distribution and Associated EM Algorithm Implementation

Abstract

This thesis introduces a new bivariate cure rate model and develops an ExpectationMaximization (EM) algorithm in R to fit the model. Within survival analysis, cure rate models describe scenarios wherein part of the population is cured and therefore would never experience the event of interest. Under this set-up, bivariate cure rate models are needed when there is a pair of events of interest. Here, a Moran-Downton bivariate Weibull distribution is used to model the paired event times of the susceptible individuals. An EM algorithm is developed here and implemented in R for this parametric bivariate cure rate model. Simulation studies are then performed to evaluate the performance of the developed model-fitting methods and finally the algorithm is applied to a real life dataset on diabetic retinopathy.