Modelos log-simétricos com fração de cura

Abstract

Long-term models are of great interest in statistical modeling that involves time-to-event data in which a fraction of the population is immune to this event. For these models, also known as cure fraction models, there are in the literature several proposals considering parametric aproach. We propose and study properties of the long-term model considering that the distributions of lifetimes of the susceptible individuals belong to the logsymmetric class of distributions. This class is characterized by continuous, strictly positive and asymmetric distributions including distributions such as log-t-Student, log-logistic I, log-logistic II, log-normal-contaminated, log-exponential-power, log-slash, among others. The log-symmetric class is quite exible to include bimodal distributions and t dataset with outlying observations. In this model, here called the log-symmetric model with cure rate, the explanatory variables are included through the parameter associated with the cure fraction. We evaluate the performance of the proposed model through extensive simulation studies and consider an application to real data in a study to identify factors which in uence the immunity of leprosy reactions in patients with leprosy.