計量生物学
Online ISSN : 2185-6494
Print ISSN : 0918-4430
最新号
選択された号の論文の9件中1~9を表示しています
原著
  • Nobuya Hayashi, Yohji Itoh
    2018 年 38 巻 2 号 p. 79-92
    発行日: 2018/03/01
    公開日: 2018/05/18
    ジャーナル フリー

    In planning a multi-regional clinical trial including Japan, the sample size for Japanese patients is often considered based on the probability of obtaining a consistent result between Japanese subpopulation and the overall study population, as recommended in the Japanese guideline “Basic Principles on Global Clinical Trials.” We review the commonly used existing method for Japanese sample size calculation based on obtaining a consistent result for survival endpoint. Through simulations, we note that Japanese sample size based on the existing method tended to have less actual power than the nominal power for consistency, especially when there is a large treatment effect. We propose alternative methods based on the delta method and numerical integrations. Our proposed methods give similar Japanese sample sizes, and our simulation studies show that our proposed methods provide the actual power close to the nominal power for consistency.

総説
  • 根本 明日香, 牛嶋 大
    2018 年 38 巻 2 号 p. 93-105
    発行日: 2018/03/01
    公開日: 2018/05/18
    ジャーナル フリー

    Quantitative high throughput sreening (qHTS) is a technique which has originally developed as a powerful tool for drug discovery and lately is expanding its application to the neighboring field, e.g. toxicological screening test for environmental chemicals. A wide variety of in-vitro biological activity of a large amount of chemical materials can be assayed with a low cost and in a short time period. As a result of two largescale pharmacogenomic studies being published in 2012, the reproducibility of the result of screening assay of cytotoxicity for 15 drugs in 471 cell lines was revealed to be unexpectedly low. The necessity of developments of statistical methods suitable for qHTS data were emphasized. In this review, the authors explain 3 statistical methods with applications to qHTS data, which has been proposed since 2013: 1. robust ridge regression estimators for nonlinear models in the purpose of testing bioactivity of chemicals; 2. Bayesian hierarchical dose-response modeling; 3. using weighted entropy to rank chemicals. Characteristics of each method were compared, and the prospects were presented.

特集
feedback
Top