论文部分内容阅读
目的：终检数据的线性回归分析问题迄今尚未解决，本文就这一问题作以探讨。方法：首先按Ｋａｐｌａｎ－Ｍｅｉｅｒ或Ｂｅｒｋｓｏｎ－Ｇａｇｅ法估计生存率，然后就生存率和其对应时间点进行线性回归分析。由此估计队列半数生存期，直接置信限，两队列半数生存期差别的显著性检验。结果：与非参数方法相比，该置信限较窄，该检验较灵敏。在给定条件下，该检验还原为两回归系数差别的显著性检验。两个工作实例：白血病和食管癌临床试验资料被用来描述其临床应用。结论：这种方法具有一定理论意义和实用价值。
Purpose: The problem of linear regression analysis of final inspection data has not been solved so far. This paper discusses this issue. Methods: Firstly, the survival rate was estimated by Kaplan-Meier or Berkson-Gage method. Then linear regression analysis was performed on the survival rate and its corresponding time points. This estimates the half-life of the cohort, the direct confidence limit, and the significance of the difference in the half-life of the two cohorts. Results: Compared with the non-parametric method, the confidence limit is narrower and the test is more sensitive. Under given conditions, the test reverts to a significant test of the difference between the two regression coefficients. Two working examples: Leukemia and esophageal cancer clinical trial data were used to describe its clinical application. Conclusion: This method has a certain theoretical significance and practical value.