纵向删失数据下线性模型的相合估计

The Consistent Estimators in Linear Model with Longitudinal Censored Data

医药临床试验,生存分析,可靠性统计等研究领域,由于考虑到时间和费用问题,研究往往有一定期限.因为研究到期的被迫结束或者某些病人中途退出试验,最后得到的试验结果往往是删失数据.对于删失数据,采用无偏转换的方法处理,方法的最大优点是得到的估计量为显式解.首先讨论了在纵向右删失数据下线性回归模型回归系数估计的均方相合性,并且把结论推广到了污染线性模型,得到了污染系数、回归系数的强相合估计.

In medical studies, survival analysis, realibity studies and many other fields, the study always has deadline because of time or cost considerations. The observations are censored data because the study is terminated or some patients withdraw from the study. In this paper, we use unbiased transformation method to deal with censored data, and the advantage of this method is that the estimator is explicit. We study the mean square consistency for the estimators of the parameters in linear regression model with longitudinal censored data. Then the result is extended to the contaminated linear model. The estimators of contaminated coefficient and regression parameter are established and the consistency of them are obtained too.