STATISTICAL EVALUATIONS OF DISSOLUTION SIMILARITY

Abstract

Statistical properties of several criteria for assessment of similarity be- tween two dissolution profiles are investigated. These include the similarity factor f2, and metrics based on the mean squared distance and the mean absolute differ- ence. The probability density function of f2 and its first two moments are derived under the assumption of multivariate normality, with special attention to compound symmetry covariance structure. The intractable nature of the distribution of f2 is demonstrated. Empirical results from a large simulation study are also presented. Advantages and drawbacks of proceduces based on the mean absolute difference and mean squared distance are discussed.