Methods for bayesian classification and prediction of survival data of long-term using modeling partition

• Grant number: 11/09454-9
• Support Opportunities: Regular Research Grants
• Start date: August 01, 2011
• End date: July 31, 2013
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Applied Probability and Statistics
• Principal Investigator: Vera Lucia Damasceno Tomazella
• Grantee: Vera Lucia Damasceno Tomazella
• Host Institution: Centro de Ciências Exatas e de Tecnologia (CCET). Universidade Federal de São Carlos (UFSCAR). São Carlos , SP, Brazil

Abstract

Models to analyze survival data assume that everyone in the study population are suceptível for the event of interest and eventually experience this event if monitoring is sufficiently long. However, there are situations where a fraction of individuals are not expected to experience the event of interest that are cured or insuceptível. An appropriate and commonly used for this is the model of mixing of Bergson and Gage (1952) also known as a model for cure rate. This model has been extensively discussed in statistical literature, yet it has many problems which are discussed in Chen et al. (1999). They developed a Bayesian model for cure rate alternative which in contrast to the model of Bergson and Gage is computationally attractive and has an intuitive interpretation and maintains the proportional hazards structure with the covariates. In this project we intend to consider a model which is an extension of the model of Chen et al. (1999), which, for flexible models the effect of covariates locally using Product Partition Model (MPP) proposed by Harting (1990) and Bayesian Partition Model (MPB) of Holmes et al. (1999). Application of this theory appears in several areas such as in Finance, Biology, Engineering, Economics and Medicine. (AU)