Characterization and classification of medicinal plant extracts according to their antioxidant activity using high-performance liquid chromatography and multivariate analysis

Abstract

Fuzzy divisive hierarchical clustering (FDHC) alongside with principal component analysis, hierarchical cluster analysis and linear discriminant analysis are efficiently employed for the characterization and clustering of some medicinal plants according to their antioxidant capacity. These methods are applied to the numerical data obtained from the chromatographic profiles monitored at 242, 260, 280, 320, 340 and 380 nm by high-performance liquid chromatography with a multistep isocratic and gradient elution system and diode array detection (HPLC-DAD). The samples were successfully classified according to the antioxidant activity determined using the DPPH assay. A correct classification rate of 100% was obtained when the samples were divided into two groups corresponding to high antioxidant activity and low antioxidant activity. Moreover, it is suggested to use the scores obtained applying principal component analysis and unprocessed data (the processed data by scaling and normalization did not improve the results), the analysis being faster with the same results. The proposed methodology could be considered as a promising tool with future applications in plant material investigations and other analytical fields.