569 – Modeling of Survival Data
The Bi-Epsilon Skew Exponential Power (BIESEP) ROC Curve
Hassan Elsalloukh
University of Arkansas at Little Rock
Ahmad Flaih
University of Arkansas at Little Rock
A new model of the Receiver Operating Characteristic (ROC) curve, the Bi-Epsilon Skew Exponential Power (BIESEP) ROC curve, is proposed in this paper. This model is a generalization of the Epsilon Skew Binormal ROC curve. Elsalloukh (2004 and 2008) provided a flexible model, the Epsilon Skew Exponential Power (ESEP), which can be adopted to accommodate asymmetry and kurtosis (platykurtic or leptokurtic) tails. The ESEP model is an appropriate choice to increase the robustness of data analysis. We develop the binormal ROC curve with a diagnostic test outcome distributed according to the ESEP model. More specifically, we derive the BIESEP ROC accuracy function. Also, we consider the estimation of BIESEP ROC curve and accuracy of a diagnostic test.
