On assessing bioequivalence using genomic data with model misspecification

This study considers the effect of a nonlinear relationship between pharmacokinetic (PK) and genomic data on bioequivalence assessment using genomic data. A true relationship between PK and genomic data is assumed to be quadratic, but a linear relationship is used instead to assess bioequivalence. Based on this misspecified model, we compare the corresponding power function and sample size for testing average bioequivalence (ABE) to those based on the true model. A numerical study is performed to explore the effect of the misspecification.