none

• Main Authors: Ya-Cheng Lee, 李亞晟
• Other Authors: Yuh-Ing Chen
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/6xf7r3

碩士 === 國立中央大學 === 統計研究所 === 106 === The major purpose of a phase I clinical trial is to estimate the maximum tolerated dose (MTD) of the drug at which the probability of the dose-limiting toxicity (DLT) is closest the targeted toxicity probability (TTP). Meanwhile, it is important to avoid assigning patients to less therapeutic doses and protect patients from overdosing in the dose escalation procedure. This article considers constructing the dose-escalation procedure based on the posterior medians of the MTD. When the maximum number of patients is reached, the MTD is then recommended by taking inference of the posterior median and mode of the MTD. Since the design involves both the posterior median and mode of the MTD, it is denoted by MEMO. The working doses for the MEMO design are calibrated based on the empirical power model for the dose-toxicity relationship. Moreover, the proposed design can be generalized to be TITE-MEMO for late-onset toxicity when data of time-to-event are available. A simulation study is finally conducted to compare the relative performance of the proposed designs to some competitive designs, for example, modified toxicity probability interval method, continual reassessment method and modified escalation with overdose control, dose assignment and MTD selection under a variety of scenarios of dose-toxicity relationship. The proposed designs generally produce a dose assignment that appropriately reduce the risk of overdose and give a recommendation of feasible MTD for the further phase II trial.