多维标度判别分析及其在医学中应用——因子分析在医学中应用

• 作 者: 陈彬 付梅菁
• 机 构: 华西医科大学公共卫生学院,四川省运动技术学院数学教研室,
• 出 处: 中国卫生统计
• 发表日期: 1995年01期
• 关键词: 判别分析 百分位数 因子分析 多维标度方程法 漏诊率 标度法 参考值范围 符合率 误诊率 标度值

判别分析法比较多,但主要问题是判别符合率低(即误诊率、漏诊率高)。某些错判对临床处理带来一定困难,或根本是不容许的。同一疾病的不同患者临床表现不同,这提示我们可应用不同“有特定医学意义的综合指标”来判断疾病。对误诊率、漏诊率高或变量间相关程度较高的资料,建议用多维标度判别分析。该法是确定医学参考值范围的多维标度方程法的推广。文中对有关问题作了讨论。

There are many discriminant analysis methods, but the main problem is that the discriminant rate is low (ie misdiagnosis rate and high rate of missed diagnosis). Some misjudgments pose some difficulties for clinical treatment, or simply do not allow it. Different patients with the same disease have different clinical manifestations, which suggests that we can use different “comprehensive indicators with specific medical significance” to judge the disease. For the data of misdiagnosis rate, high rate of missed diagnosis or high correlation between variables, multidimensional scaling discriminant analysis is recommended. This method is an extension of the multidimensional scaling equation method that defines the range of medical reference values. The relevant issues were discussed in the article.