Kairisto V, Poola A
Department of Clinical Chemistry, University of Turku, Finland.
Scand J Clin Lab Invest Suppl. 1995;222:43-60. doi: 10.3109/00365519509088450.
GraphROC for Windows is a program for clinical test evaluation. It was designed for the handling of large datasets obtained from clinical laboratory databases. In the user interface, graphical and numerical presentations are combined. For simplicity, numerical data is not shown unless requested. Relevant numbers can be "picked up" from the graph by simple mouse operations. Reference distributions can be displayed by using automatically optimized bin widths. Any percentile of the distribution with corresponding confidence limits can be chosen for display. In sensitivity-specificity analysis, both illness- and health-related distributions are shown in the same graph. The following data for any cutoff limit can be shown in a separate click window: clinical sensitivity and specificity with corresponding confidence limits, positive and negative likelihood ratios, positive and negative predictive values and efficiency. Predictive values and clinical efficiency of the cutoff limit can be updated for any prior probability of disease. Receiver Operating Characteristics (ROC) curves can be generated and combined into the same graph for comparison of several different tests. The area under the curve with corresponding confidence interval is calculated for each ROC curve. Numerical results of analyses and graphs can be printed or exported to other Microsoft Windows programs. GraphROC for Windows also employs a new method, developed by us, for the indirect estimation of health-related limits and change limits from mixed distributions of clinical laboratory data.
适用于Windows的GraphROC是一款用于临床检验评估的程序。它旨在处理从临床实验室数据库获取的大型数据集。在用户界面中,图形展示与数值展示相结合。为简单起见,除非有要求,否则不会显示数值数据。通过简单的鼠标操作即可从图形中“获取”相关数字。可以使用自动优化的组距来显示参考分布。可以选择显示具有相应置信区间的分布的任何百分位数。在敏感性-特异性分析中,与疾病和健康相关的分布会显示在同一张图中。对于任何截断限,以下数据可以在单独的点击窗口中显示:临床敏感性和特异性及其相应的置信区间、阳性和阴性似然比、阳性和阴性预测值以及效率。对于任何疾病的先验概率,都可以更新截断限的预测值和临床效率。可以生成接受者操作特征(ROC)曲线并将其组合到同一张图中,以比较几种不同的检验。会为每条ROC曲线计算具有相应置信区间的曲线下面积。分析的数值结果和图形可以打印或导出到其他Microsoft Windows程序中。适用于Windows的GraphROC还采用了我们开发的一种新方法,用于从临床实验室数据的混合分布中间接估计与健康相关的限值和变化限值。