Evaluation of macro and micronutrient elements content from soft drinks using principal component analysis and Kohonen self-organizing maps. (1st February 2019)

• Record Type: Journal Article
• Authors: Silva, Emanuela dos Santos; da Silva, Erik Galvão Paranhos; Silva, Daniélen dos Santos; Novaes, Cleber Galvão; Amorim, Fábio Alan Carqueija; dos Santos, Márcio José Silva; Bezerra, Marcos Almeida
• Is Part Of: Food chemistry. Volume 273(2019)
• Page Start: 9
• Page End: 14
• Publication Date: 2019-02-01
• Subjects: Soft drinks -- Kohonen maps -- Neural network -- Principal component analysis -- Elemental analysis -- ICP OES; Food -- Analysis -- Periodicals; Food -- Composition -- Periodicals; 664
• DOI: 10.1016/j.foodchem.2018.06.021
• Languages: English
• ISSNs: 0308-8146

Abstract

Highlights: Brazilian soft drinks could be differenced by chemometric tools. PCA and ANN were compared in the pattern recognition capacity. PCA and ANN were able to extract relevant information from soft drink samples. Abstract: This study approaches the determination of nine elements from Brazilian carbonated soft drinks of several flavors and manufactures using inductively coupled plasma optical emission spectrometry (ICP OES). The concentrations of the elements varied as follows: (in µg L −1 : Cu: 4.00–78.0; Fe: 74.0–506; Mn: 20.0–66.0; Zn: 104–584) and (in mg L −1 : Ca: 4.81–16.2; K: 6.73–260; Na: 26.0–175; S: 1.43–5.41; P: 0.186–219). Principal component analysis has shown some tendencies to form two groups according to the drink flavor (orange and cola), but only cola presented a clear and complete separation. Using Kohonen maps, it was observed a tendency to form three flavor groups: (i) cola, (ii) orange and lemon, and (iii) guarana. However, this last tool proved to be more accurate in the groups' formation.