参数竞争风险分位数模型
Parametric competing risks quantile models
投稿时间:2022-03-16  修订日期:2022-03-16
DOI:
中文关键词: 参数化  分位数回归  竞争风险
英文关键词: parametric  quantile regression  competing risks
基金项目:
作者单位邮编
胡军浩 中南民族大学数学与统计学学院 
黄荧 中南民族大学数学与统计学学院 430074
摘要点击次数: 34
全文下载次数: 0
中文摘要:
      竞争风险分位数模型[3]通过最小化不连续的L1型凸函数来进行系数推断,其可能会导致多根问题,且结果难以解释. 本文引入线性规格的参数函数,构建了一个新的参数竞争风险分位数模型. 利用积分损失最小化方法进行参数估计,解决了解不唯一性问题,极大地提高了估计效率,获得了分位数与估计系数之间的更多信息. 对估计量的一致性和渐近正态性进行了分析. 结合累积发生函数的性质,给出了模型选择与模型评估过程. 通过数值模拟说明了在标准误的意义下,参数竞争风险分位数模型更有效. 最后将模型应用到滤泡细胞淋巴瘤的临床研究中.
英文摘要:
      The competing risks quantile model[3] performs coefficient inference by minimizing discontinuous L1-type convex functions, which may lead to multiple roots and the results are difficult to interpret. In this paper, a parametric function of the linear specification is introduced to construct a new parametric competing risks quantile model. Using the integral loss minimization method for parameter estimation, the problem of non-unique results is solved, the estimation efficiency is greatly improved, and more information between quantiles and estimated coefficients is obtained. The consistency and asymptotic normality of the estimators are analyzed. Combined with the properties of cumulative incidence function, the process of model selection and model evaluation is given. Numerical simulation shows that under the standard error, the competing risks quantile model under the parameter structure is more effective. Finally, the model was applied to the clinical research of follicular cell lymphoma.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭