限制平均生存时间回归模型在生存分析中的应用

Application of restricted mean survival time regression model to survival time analysis

目的探讨限制平均生存时间(restricted mean survival time,RMST)回归模型在生存数据分析中的应用。方法运用伪值估计方法对医学数据进行限制平均生存时间回归模型实例分析,并与常见生存分析模型进行比较。结果RMST回归模型无特定模型假设,适用于不满足比例风险假定的生存数据;实例分析显示,RMST模型构建灵活,可通过设定多个τ值在多个时间段内进行估计;犯第一类错误的概率低于Cox比例风险模型,模型估算结果容易解释,能够提供在临床实践中更为实用的结论。结论在不满足比例风险假定且生存曲线有较大交叉的情形下,限制性平均生存时间模型能够提供稳定有效且易于解释的效应估计,在生存分析领域具有优良的适用性,可以作为Cox比例风险模型分析结果的补充。

Objective To explore the application of restricted mean survival time regression model to analysis of survival data.Methods Pseudo-value regression method was used to implement restricted mean survival time regression in real medical data,and then a comparison with common survival analysis models was conducted.Results The restricted mean survival time regression model without specific hypothesis was applicable to the survival data which did not meet the assumption of proportional hazards.Real case analysis displayed that the restricted mean survival time model was flexible.By setting different tau values,mean survival time could be estimated in different time periods.The results gave an intuitive explanation and provided practical conclusions in clinical practice.Conclusions In the case that the proportional risk hypothesis is not satisfied and the survival curves cross,the restricted mean survival time model can provide stable,effective estimation and straightforward interpretation,which illuminates effective applicability in the field of survival analysis and can be used as a supplement to analysis results of the Cox proportional risk model.