基于加速失效时间模型模拟生存时间应用于检测生存性状位点

对于考察预指定情形下的统计模型的性能、性质及适应性,模拟研究是非常重要的统计工具.作为生存分析中两个最受欢迎的模型之一,由于加速失效时间模型中的因变量是生存时间的对数,且此模型能够以线性形式回归带有易解释的参数的协变量,从而加速失效模型比COX比例风险模型更便于拟合生存数据.首先提出了关于带有广义F-分布的加速失效模型的模拟研究中生成生存时间的方法,然后给出了描述加速失效时间模型的误差分布和相应的生存时间之间的一般的关系式,并给出了广义F-分布是如何生成生存时间的.最后,为证实所建议模拟技术的性能和有效性,将此方法应用于检测生存性状位点的模型中.

Simulation of Survival Times Using Aft Model with Application to Mapping Survival Trait Loci

Simulation studies are an important statistical tool to investigate the performance,properties and adequacy of statistical models in the pre-specified situations.As the two most popular models in survival analysis,the accelerate failure time model(AFT) can simpler fit survival data than the Cox's proportional hazards model(PHM) main because in the AFT model,the responses are the logarithm of survival times and are linearly regressed on covariates with easily interpretable parameters.In this paper,we present the techniques to generate survival times for simulation studies regarding AFT model with the generalized F distribution and derive a general formula describing the relation between the error distributions and the corresponding survival time of the AFT model.We describe how the flexible generalized F distribution including commonly used error distributions,for instance,the exponential,the Weibull,the normal,the log-Gamma and the logistic can be applied to generate appropriate survival times.Simulation studies investigating the mapping performance of the developed method on detecting survival trait loci are conducted to demonstrate the proposed simulation techniques.