Modelagem de dados de sobrevivência via modelo de risco logístico generalizado

• Ano de defesa: 2005
• Autor(a) principal: Cremasco, Caroline Pires
• Orientador(a): Louzada Neto, Francisco
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso embargado
• Idioma: por
• Instituição de defesa: Universidade Federal de São Carlos
• Programa de Pós-Graduação: Programa de Pós-Graduação em Estatística - PPGEs
• País: BR
• Palavras-chave em Português: Estatística matemática; Análise de sobrevivência (Biometria); Modelo de Cox; Função de risco; Regressão logística
• Área do conhecimento CNPq: CIENCIAS EXATAS E DA TERRA::PROBABILIDADE E ESTATISTICA::ESTATISTICA
• Link de acesso: https://repositorio.ufscar.br/handle/20.500.14289/4515

The modeling of data of survival with the presence of covariáveis by means of the risk function has been each used time more had the easiness of interpretation One of the examples most important of risk models is the model of proportional risks considered by Cox (1972). However, this model assumes the proportionality enters the risk functions in diLerent levels of the covariáveis. To accomodate situations where the model of proportional risks is not adjusted, some types of notproportional models are being developed, as the model of sped up imperfection, considered for Prentice (1978), the model of hybrid risk of Etezadi-Amoli and Ciampi (1987) and the generalized models of hybrid risk of Louzada-Neto (1997 and 1999). In this work we explore an one new family parametric of model of dependent not-proportional risk of the time (McKenzie, 1999). This model is based on the generalization of the usual logistic function and is motivated, in part, for the necessity of if considering the eLect of the time in the modeling, and, in part, for the preference in if considering a parametric structure for the risk function. Some inferenciais procedures related this new family of models are presented.The modeling of data of survival with the presence of covariáveis by means of the risk function has been each used time more had the easiness of interpretation One of the examples most important of risk models is the model of proportional risks considered by Cox (1972). However, this model assumes the proportionality enters the risk functions in diLerent levels of the covariáveis. To accomodate situations where the model of proportional risks is not adjusted, some types of notproportional models are being developed, as the model of sped up imperfection, considered for Prentice (1978), the model of hybrid risk of Etezadi-Amoli and Ciampi (1987) and the generalized models of hybrid risk of Louzada-Neto (1997 and 1999). In this work we explore an one new family parametric of model of dependent not-proportional risk of the time (McKenzie, 1999). This model is based on the generalization of the usual logistic function and is motivated, in part, for the necessity of if considering the eLect of the time in the modeling, and, in part, for the preference in if considering a parametric structure for the risk function. Some inferenciais procedures related this new family of models are presented.