|
When comparing the dissolution data of a test product to a
reference product, the goal is to assess the similarity between
mean percent dissolution dissolved at the observed sample
time points. Several criteria for assessment of similarity
between two dissolution profiles are investigated. There are
similarity factor f2, and metrics based on the mean square
distance and the mean absolute difference. Statistical
issues of similarity factor include complex and
intractable nature of the distribution for f2, and liberality in
concluding similarity. This research conducts several simulation
for comparing f2 to other metrics based on mean squared
distance. We also examine the concept of mean absolute
difference and propose an alternative method g factor,
for assessment of the similarity. We apply Bootstrap technique
to obtain its sampling distribution. A simulation is conducted
to compare the empirical size and power between f2 and g
factor.
Keywords: Dissolution, Similarity, Bootstrap technique, Power,
Absolute Mean Difference
|