Cure Fraction Model for Interval Censoring with a Change Point based on a Covariate Threshold

In this paper, a cure fraction model for interval-censored data with a change point according to a covariate threshold is proposed. Maximum likelihood estimators of the model parameters are obtained using the Expectation Maximization (EM) algorithm. A critical challenge to this method was that the likelihood function is not differentiable with respect to the unknown change point parameter. Simulation studies were conducted to evaluate the performance of the proposed estimation method. The numerical results showed that the new model represents a valuable advancement of cure models.