A2057
Title: A copula-based defective Gompertz cure rate model for dependent censoring
Authors: Silvana Schneider - Federal University of Rio Grande do Sul (Brazil) [presenting]
Abstract: Traditional survival models are often inadequate for analyzing datasets that include a cure fraction, especially in the case of competing risks. One of the challenges in this analysis is the presence of dependent censoring, which occurs when the censoring times are not independent of lifetimes. Such censoring can occur when patients leave follow-up due to either clinical improvement or disease progression. In such cases, the assumption of independent censoring, commonly adopted in traditional models, may be violated, impacting survival estimates and statistical inference. To address this limitation, a copula-based regression model for dependent censoring is proposed, incorporating the defective Gompertz distribution to model the cure rate. The proposed model employs the Clayton copula function to capture possible dependence structures between lifetimes and censoring times. The proposed methodology is validated through simulation studies and an application to a dataset that has not yet been explored in the literature. Specifically, the model is applied to breast Cancer data obtained from the Hospital Cancer Registry of the state of Paraiba, Brazil, showcasing its ability to capture complex dependencies in survival outcomes and to estimate the cure fraction within this population.
