[Fitting of ROC curves for continuous data by using correction of the mean and standard deviation].

ABSTRACT

We examined ROC curve fitting with continuous rating data by using a corrected mean value and corrected standard deviation as binormal ROC parameters. Curve fitting was performed by using corrected categorical data in the same way as the continuous data. A large residual of curve fitting was found in cases of difference between means of positive and negative signal data greater than three times the standard deviation, or with a ratio of the standard deviation of positive and negative signal data above six. We conclude that our proposed method performed well for ROC curve fitting except in some special cases.