Hayashi C
Gan To Kagaku Ryoho. 1986 Apr;13(4 Pt 2):1149-55.
The subjects which are often encountered in the statistical design and analysis of data in medical science studies were discussed. The five topics examined were: Medical science and statistical methods So-called mathematical statistics and medical science Fundamentals of cross-tabulation analysis of statistical data and inference Exploratory study by multidimensional data analyses Optimal process control of individual, medical science and informatics of statistical data In I, the author's statistico-mathematical idea is characterized as the analysis of phenomena by statistical data. This is closely related to the logic, methodology and philosophy of science. This statistical concept and method are based on operational and pragmatic ideas. Self-examination of mathematical statistics is particularly focused in II and III. In II, the effectiveness of experimental design and statistical testing is thoroughly examined with regard to the study of medical science, and the limitation of its application is discussed. In III the apparent paradox of analysis of cross-tabulation of statistical data and statistical inference is shown. This is due to the operation of a simple two- or three-fold cross-tabulation analysis of (more than two or three) multidimensional data, apart from the sophisticated statistical test theory of association. In IV, the necessity of informatics of multidimensional data analysis in medical science is stressed. In V, the following point is discussed. The essential point of clinical trials is that they are not based on any simple statistical test in a traditional experimental design but on the optimal process control of individuals in the information space of the body and mind, which is based on a knowledge of medical science and the informatics of multidimensional statistical data analysis.
讨论了医学科学研究中数据统计设计和分析中经常遇到的主题。所考察的五个主题分别是：医学科学与统计方法；所谓的数理统计与医学科学；统计数据交叉表分析及推断的基础；多维数据分析的探索性研究；个体的最优过程控制、医学科学与统计数据信息学。在第一部分，作者的统计数学思想被描述为通过统计数据对现象进行分析。这与科学的逻辑、方法论和哲学密切相关。这种统计概念和方法基于操作性和实用性思想。在第二部分和第三部分特别关注对数理统计的自我审视。在第二部分，就医学科学研究对实验设计和统计检验的有效性进行了深入考察，并讨论了其应用的局限性。在第三部分，展示了统计数据交叉表分析和统计推断中明显的悖论。这是由于对（多于两个或三个）多维数据进行简单的两重或三重交叉表分析的操作，而没有考虑复杂的关联统计检验理论。在第四部分，强调了医学科学中多维数据分析信息学的必要性。在第五部分，讨论了以下要点。临床试验的要点在于，它们并非基于传统实验设计中的任何简单统计检验，而是基于身心信息空间中个体的最优过程控制，这是基于医学科学知识和多维统计数据分析信息学的。
