對於一種診斷的準確性，通常利用ROC曲線下的面積來評估，而目前的醫學診斷中，利用多個診斷方法同時評估某種疾病是常見的，例如: 生物微晶片(microarray)上數個基因的表現，同時用來診斷某個疾病。針對同時使用 種不同的診斷方法去評估某種疾病時，若將此 個ROC曲線下面積分別以算術平均、及以幾何平均加權，本論文主要是探討及比較以此兩種方法所得的測度量(沒有考慮到診斷方法相關性)與Ma & Ou (2003)所提出的多變量ROC曲線之整合測度量(有考慮到診斷方法相關性)有何不同。並將這些方法運用到生物微晶片上，評估多個基因同時使用來診斷疾病的準確性，本文採用一組著色性乾皮症患者和正常人的基因表現資料來說明，最後利用模擬生成的資料針對準確性作進一步性質的探討。

　　An important statistical tool for describing diagnostic accuracy is the receiver operating characteristic (ROC) curve. There is an increasing use of ROC curve for assessing the effectiveness of continuous diagnostic markers in diagnosis of patient. However, scarce literature exists for evaluation of ROC curve based on multiple continuous markers.
　　Two methods for estimation of the area under ROC curves based on multiple markers are proposed. These two methods are based on arithmetic and geometric mean of areas under ROC of individual markers. A numerical example using the data from a microarray experiment illustrates the applications of the proposed methods. Finally, a simulation was conducted to evaluate the performance of the proposed methods.

References

1.Bamber, D., “The area above the ordin aldominance graph and the area below the receiver operation graph”, J. Math. Psych. 12, 387-415, 1975.
2.Chang, H. C., Tsai, J. H., Guo, Y. L., Huang, Y. H., Tsai, H. N., Tsai, P. C., Huang, W., “Differential UVC-induced expression of the growth arrest and DNA damage-inucible gene gadd45 in xeroderma pigmentosum C cells identified by cDNA
microarray”, submitted, 2003.
3.DeLong, E., DeLong, D., and Clarke-Pearson, D., “Comparing the areas under two or more correlated receiver operation characteristic curves : A nonparametric approach”, Biometrics, 44, 837-845, 1988.
4.Efron, B., “Bootstrap methods : another look at the jacknife”,Annals Statist., 7, 1:26, 1979.
5.Guo, Y. L., Chang, H. C., Tsai, R. H., Huang, J. C., Li, C., Young, K. C., Wu, L. W., Lai, M. D., Liu, H. S., Hung, W., “Two UVC-induced stress response pathways in HeLa cells identified by cDNA microarray”, Environ Mol Mutagan, 40, 122-128, 2002.
6.Hanley, J.A. and McNeil, B.J., “The meaning and use of the area under of receiver operating characteristic (ROC) curve”, Radiology,143, 29-36, 1982.
7.Hanley, J. A. and McNeil, B. J., “A method of comparing the area under two ROC curves derived from the same cases”, Radiology, 148, 839-843, 1983.
8.Hollander, M. & Wolfe, D. A., “Nonparametric statistical methods”, Wilty, New York, 1973
9.Ma, M. C. and Ou, J. C., “The eatimation of the area under multivariate ROC Curve”, submitted, 2003.
10.Ma, M. C. and Zhuang, J. R., “On the Evaluation of Different Statistical Procedures for Microarray Data”, submitted, 2003.