论文部分内容阅读
由正态近似法产生的生存率置信限，在给走样本当生存率逐渐接近０或１时，可出现置信限超出（０，１）区间的不合理现象。本文将Ｋａｌｂｆｌｅｉｓｃｈ和Ｐｒｅｎｔｉｃｅ的参数转换方法与同源性生存率方差估计值结合起来构造一种改良生存率置信限，避免了这种不合理现象．并改进了正态逼近的充分性。附有实例描述其临床应用。
The confidence limit of survival rate generated by the normal approximation method is that when the survival rate gradually approaches 0 or 1 in the sample given away, the unreasonable phenomenon that the confidence limit exceeds the range of (0, 1) can occur. This paper combines Kalbfleisch and Prentice’s parametric transformation methods with homogeneity variance estimates to construct a modified survival confidence limit to avoid this unreasonable phenomenon. And improve the adequacy of normal approximation. With examples to describe its clinical application.