Some Applied Aspects Of Multivariate Parallel-line Assays (selectivity Ratio, Marginal Parallelism, Relative Potency).

Abstract

This dissertation proposes analytical techniques for the analysis of correlated multivariate parallel-line assays. The condition of marginal parellelism which is essential for the validity of a multivariate parallel-line assay is investigated. Combined estimates of the common slope parameters are proposed along with the proof that they are the maximum likelihood estimates. Two equivalent tests are proposed for the constancy of the relative potency over time. Also an unbiased minimum variance combined estimate of the relative potency is introduced. Next, this dissertation proposes a selectivity ratio test for comparing both time-dependent and time-independent relative potencies and a double selectivity ratio test for comparing selectivity ratios of time-dependent relative potencies. The selectivity ratio test makes it possible to determine the magnitude of the relative potency of a new drug as compared to a standard drug for both the intended therapeutic effect and side-effects for two or more physiological systems. The double selectivity ratio test investigates the nature of selectivity ratios obtained for time-dependent relative potencies.