- 作者: 周若珍; 謝復興
- 作者服務機構: 國立清華大學應用數學研究所
- 中文摘要: 本文討論兩種藥品在臨床試驗中的比較法。本文不假設觀測值的變異數為已知,亦不要求每種藥品有相同數目的病人來做實驗,因而提出的比較法兼顧了道德問題及合理性。
- 英文摘要: For the problem of comparing two treatments in clinical trials, procedures proposed by Anscombe[1], Begg and Metha [2] and Chernoff and Petkau [3] are based on paired normal observations with knownvariances. In this paper, we extend the problem in two directions. Firstly, the variance is assumed to berandom with an inverse gamma prior distribution. Secondly, the patients are treated one by one with thetreatment judged as better at the time that they arrive. Procedures are proposed so that the expected riskis minimized. Critical values for stopping the experimental period are obtained. Simulation studies are alsocarried out to reveal the performances of the procedures.
- 中文關鍵字: Bayesian estimation; sequential procedure
- 英文關鍵字: --