Genetic Algorithms-Based Approach for Wavelength Selection in Spectrophotometric Determination of Vitamin B12 in Pharmaceutical Tablets by Partial Least-Squares

Abstract

This study illustrated the genetic algorithms (GA) is useful tool in wavelength selection for the spectrophotometric determination of vitamin B12 in pharmaceutical tablets by partial least-squares calibration procedure (PLS). Because of interference in the spectra, a training set of the mixtures of vitamin B12 and tablet excipients was built to be used as calibration set for modeling. The absorption spectra of all solutions were recorded each 1nm in the range 354-600nm, using 0.1N hydrochloric acid as solvent. A validation set was employed to verify the models and a commercial sample was analysed. GA-PLS model leads to 86% improvement of predictive ability for vitamin B12, when comparing with full spectrum PLS model. It is concluded that only some variables in GA-PLS modeling are relevant for prediction and removing the non-relevant variables can improve the predictive capability.