在新藥開發的臨床試驗中，會將受試的病人分成控制組(control group)和實驗組(treatment group)，給予控制組的服用舊藥或安慰劑、實驗組的服用欲開發之新藥，透過分析決定是否讓新藥上市。除了選出藥效最佳的藥外，如何控制實驗成本亦是科學家所關心的。因此，Cheng-Shiun Leu與Bruce Levin針對第二期的臨床試驗，提出了一個選取藥物的實驗程序，稱為adaptive design in phase-II clinical trials。這些研究主要是適用於服藥結果只有治療成功或失敗的藥物，但對於高血壓、高膽固醇等疾病的患者而言，其服藥反應為連續型變數。所以本文提出適用於服藥反應為常態分配下的選藥程序，無調適NDD和調適NDD，並以貝氏觀點建構成本函數為指標。利用R軟體以蒙地卡羅法進行模擬，探討在不同數量組合的新舊藥和先驗分配下，如何設定檢定的顯著水準和最多受試者人數，才能以最低的成本選出藥效最佳的藥。證實了在藥效參數服從beta分配且服藥反應為常態時，使用調適NDD較無調適NDD佳，並且找出了最佳化設定。

In the clinical trials of new drug development, patients will be divided into control group and treatment group. Let patients in control groups take old drugs or placebos and those in treatment group take new drugs. Scientists will determine whether the efficacy of new drugs are truly better through analysis. Besides selecting the drugs with best efficacy, scientists also care about controlling the experimental cost. Therefore, Cheng-Shiun Leu and Bruce Levin have designed a procedure for selecting drugs in phase-II clinical trials, which is called adaptive design in phase-II clinical trials. These researches mainly focus on the drugs whose responses are only success or failure. However, to patients with hypertension and hypercholesterolemia, their responses of taking medicine are continuous variables. In this thesis, we provide sequential procedures (non-adaptive NDD and adaptive NDD) for drugs with Normal distributed responses and define a cost function as criteria based on Bayesian. Through R language and Monte Carlo method, we study how to set the significance level and the largest number of patients to select the drugs with best efficacy using minimum costs. Finally, we verify that the adaptive NDD is better than non-adaptive NDD and solve the optimizations.

[1] Cheung, K. (2008). ‘Simple sequential boundaries for treatment selection in
multi-armed randomized clinical trials with a control’. Biometrics. 64, 940-949.
[2] Leu, C.S. and Levin, B. (1999a). ‘Proof of a lower bound formula for the expected
reward in the Levin-Robbins sequential elimination procedure’. Sequential Analysis.
18, 81-105.
[3] Leu, C.S. and Levin, B. (1999b). ‘On the probability of correct selection in the
Levin-Robbins sequential elimination procedure’. Statistica Sinica. 9, 879-891.
[4] Leu, C.S. and Levin, B. (2004). ‘Selecting the best subset of b out of c coins with the
Levin-Robbins sequential elimination procedure: Proof of the lower bound formula for
the probability of correct selection in the case b = 2, c = 4’. Technical Report #B-91,
Department of Biostatistics, Columbia University.
[5] Leu, C.S. and Levin, B. (2008). ‘A generalization of the Levin-Robbins procedure for
binomial subset selection and recruitment problems’. Statistica Sinica. 18, 203-218.
[6] Leu, C.S. and Levin, B. (2013). ‘On two lemmas used to establish a key inequality that
implies the lower bound formula for the probability of correct selection in the
Levin-Robbins-Leu family of sequential binomial subset selection procedures’. Technical
Report #B-148, Department of Biostatistics, Columbia University.
[7] Leu, C.S., Cheung, K., and Levin, B. (2011). ‘Subset selection in comparative
selection trials’. In: Bhattacharjee, M., Dhar, S.K., and Subramanian, S. (eds.) Recent
Advances in Biostatistics: False Discovery, Survival Analysis, and Other Topics.
Series in Biostatistics, 4, 271-288..
[8] Levin, B. and Robbins, H. (1981). ‘Selecting the highest probability in binomial or
multinomial trials’. Proc. Natl. Acad. Sci. USA. 78, 4663-4666.