The Geometric Mean Functional Relationship Approach to Linear Regression in Pharmaceutical Studies: Application to the Estimation of Binding Parameters
Georgia N. Valsami, Panos E. Macheras · Pharmacy and Pharmacology Communications · 1995
The aim of this study was to draw attention and apply the geometric mean functional relationship (GMFR) approach to the analysis of data which are currently analysed with the ordinary least-squares method, in spite of the fact that both variables are subject to error. The method was applied to drug-protein binding data using erroneous simulated data, generated from the Scatchard model with one class of binding sites. For the present study, a computer programme in BASIC was constructed to perform linear regression analysis by means of the “least-triangles” (LT) approach to GMFR and the ordinary least-squares (LS) method. The least-triangles approach for linear regression analysis was proven to be superior to the ordinary least-squares method when applied to contaminated simulated binding data. The method requires minimum computation and it can be applied to many other types of pharmaceutical studies where linear regression is applied and both variables are subject to error.