目前,金沙8888js官方刘祎博士及其合作者在统计学期刊《Statistical Methods in Medical Research》上发表了题为“A new joint screening method for right-censored time-to-event data with ultra-high dimensional covariates”的研究论文。
上述论文研究的是右删失生存时间的超高维变量联合筛选方法BJASS (Buckley-James assisted sure screening)。BJASS方法基于加速失效时间模型,对回归系数进行稀疏限制,考虑了变量之间的联合效应,避免了以往变量筛选方法只考虑边际相关性或者必须进行迭代的限制。刘祎讲师与合作者提出利用合成数据和Buckley-James补值数据的两步法,来实现对重要变量的筛选过程,并提出了相应的数值算法实现。该方法理论上可以以概率1选择出重要的变量,并且数值表现良好。审稿人评价:‘I commend the authors on their manuscript, which tackles the problem of variable selection inthe more difficult, and less studied, area of censored outcomes with AFT models...The authors do a nice job of justifying their approach from a theoretical perspective, and then highlight it's performance in some simulated examples.’。
该研究工作是刘祎博士与曲阜师范大学陈晓林副教授及加州大学洛杉矶分校李刚教授合作完成,刘祎博士为第一作者,本项工作得到国家自然科学基金的资助。
论文链接地址:
https://journals.sagepub.com/doi/abs/10.1177/0962280219864710