在臨床試驗中，研究者欲確保藥品的安全性及討論人體對藥物的反應，故只針對符合限制條件下的患者進行臨床試驗。在實務上，由於研究時間有限，樣本招募速度不如預期，研究者可能會變更受試者的限制條件，然而研究者感興趣的仍是原受試者限制條件範圍下患者之平均療效反應，則此時估計的統計方法可能需進行修正或調整。

本論文在受試者限制條件變更的情況下，當資料中存在右設限觀察值且療效反應為存活時間時，利用統計模式連結療效反應與限制條件的關係，以最大概似法估得模式中的參數，並進而估計出原受試者限制條件範圍下平均存活時間。最後經由統計模擬探討其準確性。

Adapted design has become more and more attractive and popular in clinical development due to its flexibility. Chow and Shao (2005) have developed a general inference procedure based on linear regression model for protocol amendments that are related to some covariates of the on-going trial, such as modifications of patient inclusion criteria. Note that their procedure can not be directly applied to the data set with right-censored observations, therefore, this thesis extends their procedure for right-censored data based on log-linear models where the distribution of survival times is known. Then the maximum likelihood estimators of survival function and mean survival time for target population are derived. The accuracy of these estimators is examined by a simulation study.

[1] Chow, S. C. and Shao, J. (2005), “Inference for clinical trials with some
protocol amendments”, Journal of Biopharmaceutical Statistics, 15(4):
659-666.
[2] Lawless, J. F. (2003), “Statistical Models and Methods for Lifetime
Data”, 2nd edition, New York: John Wiley and Sons, Hoboken.