MULTI-RESPONSE OPTIMIZATION OF METFORMIN -HYDROCHLORIDE DRUG VIA EXPERIMENTAL DESIGN TECHNIQUES APPLIED ON THE ADAPTIVE LOCAL LINEAR REGRESSION MODEL FOR RESPONSE SURFACE METHODOLOGY

Abstract

The challenge encountered by industrial statisticians and engineers during statistical process control in other to build-up and formulates pharmaceutical optimization of drugs have prompted several choices of experimental design tool and regression models. The most commonly applied regression model is the second-order polynomial which may perform poorly due to model misspecification. In this paper, we present two experimental design methods namely; the Full Factorial Design (FFD) and the Circumscribed Central Composite Design (CCCD) applied to an existing Adaptive Local Linear Regression ( model to ameliorate the problem of boundary bias for a multi-response problem. The FFD do not make reference to the star points and as such could not address variability in the data, hence we also applied the CCCD to accommodate the star points in order to maintain rotatability and curvature in the data. In the application, we minimized Metformin Hydrochloride (Met-HCL) drug usage via FFD on and CCCD on and the results from the goodness-of-fit statistics, residual plots and optimization were obtained and analyzed. The that utilized CCCD outperformed that uses FFD in terms of the goodness-of-fit statistics, minimum residual plots as it relates to the zero residual line and optimization of Met-HCL for Response Surface Methodology (RSM) data.