Bayesian reasoning and multi-state models are used to assess the progression of stage IV non-small cells lung cancer (NSCLC) through the disability model with three states: the initial one, which is determined by the stage IV pa- tient’s diagnosis time, the tumour progression and the patient’s death. Transition probabilities between states are expressed in terms of the haz- ard rate functions for times between transitions, which we analyze through Weibull accelerated failure time regression models. Uncertainty about the parameters of the model is expressed in terms of its posterior distribution and it has been propagated to the hazard rate functions of the times between transitions. We can thus obtain the posterior predictive distribution for the transition probabilities, given the time and the covariates, which offers a satisfactory description of the dynamics of the system. Data for the study comes from the Infanta Cristina Hospital of Madrid, Spain, and consist of survival times for stage IV NSCLC patients and measures of several covariates that may be related to the disease, observed from January 2008 to December 2010.
Bayesian multi-state models for assessing the progression of stage IV non small-cell lung cancer
PERRA, SILVIA;CABRAS, STEFANO;Castellanos M. E;
2011-01-01
Abstract
Bayesian reasoning and multi-state models are used to assess the progression of stage IV non-small cells lung cancer (NSCLC) through the disability model with three states: the initial one, which is determined by the stage IV pa- tient’s diagnosis time, the tumour progression and the patient’s death. Transition probabilities between states are expressed in terms of the haz- ard rate functions for times between transitions, which we analyze through Weibull accelerated failure time regression models. Uncertainty about the parameters of the model is expressed in terms of its posterior distribution and it has been propagated to the hazard rate functions of the times between transitions. We can thus obtain the posterior predictive distribution for the transition probabilities, given the time and the covariates, which offers a satisfactory description of the dynamics of the system. Data for the study comes from the Infanta Cristina Hospital of Madrid, Spain, and consist of survival times for stage IV NSCLC patients and measures of several covariates that may be related to the disease, observed from January 2008 to December 2010.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
