Bravata Dena M, Shojania Kaveh G, Olkin Ingram, Raveh Adi
Center for Primary Care and Outcomes Research, Stanford University School of Medicine, 117 Encina Commons, Stanford, CA 94305-6019, U.S.A.
Stat Med. 2008 May 30;27(12):2234-47. doi: 10.1002/sim.3078.
Many critical questions in medicine require the analysis of complex multivariate data, often from large data sets describing numerous variables for numerous subjects. In this paper, we describe CoPlot, a tool for visualizing multivariate data in medicine. CoPlot is an adaptation of multidimensional scaling (MDS) that addresses several key limitations of MDS, namely that MDS maps do not allow for visualization of both observations and variables simultaneously and that the axes on an MDS map have no inherent meaning. By addressing these issues, CoPlot facilitates rich interpretation of multivariate data. We present an example using CoPlot on a recently published data set from a systematic review describing clinical features and disease progression of children with anthrax and provide recommendations for the use of CoPlot for evaluating and interpreting other healthcare data sets.
医学中的许多关键问题都需要对复杂的多变量数据进行分析，这些数据通常来自描述众多受试者的众多变量的大型数据集。在本文中，我们描述了CoPlot，这是一种用于可视化医学多变量数据的工具。CoPlot是多维缩放（MDS）的一种改编形式，它解决了MDS的几个关键局限性，即MDS图不允许同时可视化观测值和变量，并且MDS图上的轴没有内在意义。通过解决这些问题，CoPlot有助于对多变量数据进行丰富的解释。我们展示了一个使用CoPlot的例子，该例子基于最近发表的一项系统评价数据集，该数据集描述了炭疽病儿童的临床特征和疾病进展，并为使用CoPlot评估和解释其他医疗数据集提供了建议。
