A data fusion approach combining chromatographic and spectroscopic profiles is proposed for the discrimination and classification of soothing herbs in different types of herbal preparations. Particularly, chamomile, lavender, passionflower, and valerian were considered. The proposed data fusion approach revealed a higher clusterization ability than each analytical technique in a separate way, which was assessed through an exploratory analysis based on Principal Component Analysis (PCA) coupled to Silhouette analysis: percentage of samples with a negative Silhouette width were 19, 15 and 10 for chromatography, spectroscopy and data fusion, respectively. Furthermore, a Partial Least Squares – Discriminant Analysis (PLS-DA) model developed based on data fusion was able to perfectly discriminate samples of chamomile, passionflower, and valerian in a set of 20 samples, overcoming the difficulties related to dealing with different types of herbal preparations including pure herbs, infusions, tablets, capsules and herbal drops.
Read full abstract