| Περίληψη : | Various chemometrical applications have been included in USA’s and European legislation, regarding the quality control of produced medicines. One of the most critical parameters is to be determined the quantity of active ingredient (API) in the medicine. In current thesis the objective was the quantification of API ingredient in a pharmaceutical mixture by employing the novel Bayesian Bridge (BB) method. Additionally, in order to be assessed the results from BB was employed the Partial Least Squares (PLS) method. For the model development was used data that sourced from Near Infrared (NIR) instrument. Over a 700 hundred variables (wavelengths) get measured by NIR spectrometer, per sample. In total we had 43 samples, based on balanced experimental design. The assessment of developed model was performed using the leave-one-out method as cross-validation technique and by calculating the RMSECV. Even though, both methods gave satisfactory results, capable to predict accurately the quantity of API in mixtures of medicines, the BB estimator outperforms the PLS with an average 0.13 and 0.19 RMSECV respectively.
|
|---|
Copyright © 2013 Βιβλιοθήκη Ο.Π.Α. Με την επιφύλαξη παντός δικαιώματος.