Multivariate lifetime models to evaluate long-term survivors in medical studies

Detalhes bibliográficos
Ano de defesa: 2019
Autor(a) principal: Oliveira, Ricardo Puziol de
Orientador(a): Não Informado pela instituição
Banca de defesa: Não Informado pela instituição
Tipo de documento: Tese
Tipo de acesso: Acesso aberto
Idioma: eng
Instituição de defesa: Biblioteca Digitais de Teses e Dissertações da USP
Programa de Pós-Graduação: Não Informado pela instituição
Departamento: Não Informado pela instituição
País: Não Informado pela instituição
Palavras-chave em Português:
Link de acesso: http://www.teses.usp.br/teses/disponiveis/17/17139/tde-08012020-110425/
Resumo: Multivariate survival data are presented in the literature in all shapes and sizes. A common situation is the presence of correlated lifetimes when an individual is followed-up for the occurrence of two or more types of events, or when distinct individuals have dependent event times. In many applications involving these type of data, it is common the use of continuous random variable modeling approach. In this direction, the multivariate normal distribution is the most common used since it has friendly properties such as a readily interpretable dependence structure. Moreover, in most of these studies, there is the presence of covariates such as treatments, group indicators, individual characteristics, or environmental conditions, whose relationship to lifetime is of interest. In this situation, it is needed to assume lifetime regression models. In this way, the well known Cox proportional hazards model and its variations, using the marginal hazard functions employed for the analysis of multivariate survival data in literature are not enough to explain the complete dependence structure of pair of lifetimes on the covariate vector. In this thesis, it is presented some new multivariate lifetime models assuming cure rate structure based on mixture and non-mixture approaches for the analysis of long-term survivors applied to medical studies. The proposed models could be also useful to study the dependence structure of pair of lifetimes on the covariate vector X. The results emerging from this study reinforce the fact that the search of appropriate multivariate lifetime distributions could be extremely difficult depending on the correlation structure of the lifetime data. However, the proposed methodology could be very useful in the medical lifetime data analysis where the interest is the estimation of the fraction of patients in the studied population who never experience the event of interest. In addition, the identification of important covariates was also easily obtained assuming the proposed models even using non-informative priors for the parameters of the model, under a Bayesian approach. The results could be also extended to other cross-over trials in clinical research; reliability analysis in engineering; risk analysis in economics; among many other areas. For reproducible research, the general framework for the computer codes of the proposed modeling approach is also presented which could be carried out using free R or OpenBugs free softwares.