Chemometric Tools Applied to Evaluation of Fruit Bioactive Compounds Extraction

• Published in: Food analytical methods 2020-05, Vol.13 (5), p.1176-1189
• Main Authors: da Silva Santos, Bruna Rosa, Minho, Lucas Almir Cavalcante, Silva, Emmanuelle Ferreira Requião, da Silva Sauthier, Maria Celeste, da Cruz Caldas, Jamile, da Silva, Erik Galvão Paranhos, de Andrade Santana, Débora, dos Santos, Walter Nei Lopes
• Format: Article
• Language: English
• Subjects: Analytical Chemistry; Anthocyanins; Antioxidants; Artificial neural networks; Bioactive compounds; Biological activity; Chemistry; Chemistry and Materials Science; Chemistry/Food Science; Classification; Discriminant analysis; Ellagic acid; Flavonoids; Food Science; Fruits; High performance liquid chromatography; Independent variables; Liquid chromatography; Microbiology; Multilayer perceptrons; Neural networks; Phenolic compounds; Phenols; Polyphenols; Quercetin; Rutin; Spectrophotometry; Vanillic acid
• ISSN: 1936-9751; 1936-976X
• DOI: 10.1007/s12161-020-01728-0

A previously fully validated method using high-performance liquid chromatography with diode array detection (HPLC-DAD) was applied to determine 12 bioactive phenolic compounds in 9 different fruits consumed in Salvador, Bahia, Brazil. A central composite design (CCD) was employed to investigate the effects of independent variables in the extraction method for spectrophotometric determinations. The model showed a good correlation between the predicted and experimental values. The results indicate that the fruits are rich in polyphenols, mainly ellagic acid, vanillic acid, rutin and quercetin. Total polyphenol content (TPC), total flavonoid content (TFC), total anthocyanin content (TAC), and antioxidant capacity were also determined by DPPH assay and can be related to the diverse range of phenolics detected. Kohonen Self- Organizing Map (SOM) Artificial Neural Network was applied for more insights about cluster separation and the influence of each variable considering one of its main characteristics related to the treatment of non-linear data. A true classification model partial least squares-discriminant analysis (PLS-DA) and multilayer perceptron (MLP) artificial neural network were applied to the current samples, and while MLP proved to be a suitable technique for the classification of different fruits evaluated in this study, PLS-DA obtained a very large experimental error due to the similarities of the experimental data and also due to the existence of non-linear relationships between the variables.