The impact of censored survival data on Bayesian inference is assessed when estimating Bayesian Weibull mixture models through a simulation study and an application to microarray data. The simulation study was carried out with different parameter configurations of the mixture model, that is, two well-separated components and two strongly overlapping components for data generation each with five different levels of censoring. The Bayesian approach via Markov Chain Monte Carlo was used to estimate the parameters of Weibull mixture model. The issue of label switching and model evaluation are also considered.udKeywords: Bayesian modelling, Censored data, Markov Chain Monte Carlo, mixture model, Survival analysis,Weibull distribution.
展开▼
机译:通过模拟研究和微阵列数据的应用估计贝叶斯威布尔混合模型时,评估了审查的生存数据对贝叶斯推断的影响。在混合模型的不同参数配置下进行了仿真研究,即,两个完全分开的组件和两个强烈重叠的组件用于数据生成,每个组件都有五个不同的审查级别。采用马尔可夫链蒙特卡罗方法的贝叶斯方法估计了威布尔混合模型的参数。 ud关键字:贝叶斯建模,删失数据,Markov Chain Monte Carlo,混合模型,生存分析,Weibull分布。
展开▼