The geometric power half-normal regression model with cure rate

Abstract

In this paper we consider the geometric cure rate model defined in [16], using for $S_0(\cdot)$, the survival function of carcinogenic cells, an extension of the half-normal distribution based on the distribution of the maximum of a random sample. The implementation of maximum likelihood estimation for the model parameters is discussed and, nally, the model is fitted to a real database (Melanoma data set), and comparisons are performed with alternatives to the new $S_0(\cdot)$.