Survival analysis that involves moderate/high dimensional covariates has become common. Most of the existing analyses have been focused on estimation and variable selection, using penalization and other regularization techniques. To draw more definitive conclusions, a handful of studies have also conducted inference. The recently developed mFDR (marginal false discovery rate) technique provides an alternative inference perspective and can be advantageous in multiple aspects. The existing inference studies for regularized estimation of survival data with moderate/high dimensional covariates assume the Cox and other specific models, which may not be sufficiently flexible. To tackle this problem, the analysis scope is expanded to the transformation model, which is robust and has been shown to be desirable for practical data analysis. Statistical validity is rigorously established. Two data analyses are conducted. Overall, an alternative inference approach has been developed for survival analysis with moderate/high dimensional data.

涉及中/高维协变量的生存分析已经变得很普遍。大多数现有分析都集中在估计和变量选择上，使用惩罚和其他正则化技术。为了得出更明确的结论，一些研究也进行了推断。最近开发的 mFDR（边际错误发现率）技术提供了另一种推理视角，并且在多个方面都具有优势。现有的用于对具有中/高维协变量的生存数据进行正则化估计的推理研究假设 Cox 和其他特定模型可能不够灵活。为了解决这个问题，分析范围扩展到了转换模型，该模型是健壮的，并且已被证明是实际数据分析的理想选择。严格建立统计有效性。进行了两次数据分析。总体而言，已经开发了一种替代推理方法，用于使用中/高维数据进行生存分析。