Modelo de mistura padrão com fragilidade gama: aplicação a dados de melanoma

Abstract

Cancer is a malignant disease that can affect several organs. When it spreads throughout the body and reaches other organs, it is called metastasis, and this is the leading cause of cancer-related death. Although skin cancer is the most common in Brazil, and corresponds to around 30% of all malignant tumors registered in the country, melanoma represents only 4% of malignant neoplasms, according to the National Cancer Institute (INCA ). Intense and sudden exposure to ultraviolet (UV) radiation can cause sunburn. This exposure results in a mutation in the DNA of the cells responsible for producing melanin, developing abnormal cells and in turn leading to the development of melanoma. Due to the high incidence of death from cancer, it is crucial to invest in research, education and resources for the prevention, diagnosis and effective treatment of this disease. Among the statistical techniques used to build the models, survival analysis stands out. The application of this technique in the medical field seeks to estimate, for example, the survival time of patients after diagnosis of the disease, in addition to providing information on the progression of the disease, the effectiveness of treatments and associated risk factors. Based on patient survival data, it is possible to identify patterns and determine which characteristics are associated with better or worse prognoses. In this context, this work aims to consider a cure fraction model with fragility for analyzing melanoma data. This model takes into account the possibility that some patients can be cured of the disease, while others cannot. Furthermore, it incorporates the idea of fragility, which allows taking into account the variability not captured by the explanatory variables observed in the study. The proposed methodology will be applied to a database from the Oncocentro de São Paulo Foundation (FOSP). This database contains relevant information about melanoma patients, including their survival times and clinical characteristics.