Sameera G, Vardhan R Vishnu, Sarma K V S
a Department of Statistics , Pondicherry University , Pondicherry , India.
b Department of Statistics , Sri Venkateswara University , Tirupati , India.
J Biopharm Stat. 2016;26(3):421-31. doi: 10.1080/10543406.2015.1052479. Epub 2015 May 26.
The classification scenario needs handling of more than one biomarker. The main objective of the work is to propose a multivariate receiver operating characteristic (MROC) model which linearly combines the markers to classify them into one of the two groups and also to determine an optimal cut point. Simulation studies are conducted for four sets of mean vectors and covariance matrices and a real dataset is also used to demonstrate the proposed model. Linear and quadratic discriminant analysis has also been applied to the above datasets in order to explain the ease of the proposed model. Bootstrapped estimates of the parameters of the ROC curve are also estimated.
分类场景需要处理不止一个生物标志物。这项工作的主要目标是提出一种多变量接收器操作特征（MROC）模型，该模型将这些标志物线性组合以将它们分类为两组中的一组，并确定一个最佳切点。针对四组均值向量和协方差矩阵进行了模拟研究，并且还使用了一个真实数据集来演示所提出的模型。线性和二次判别分析也已应用于上述数据集，以说明所提出模型的简便性。还估计了ROC曲线参数的自助法估计值。
