Premium
New guidelines for prediction of antioxidant activity of Lactuca sativa L. varieties based on phytochemicals content and multivariate chemometrics
Author(s) -
Jevrić Lidija R.,
Karadžić Milica Ž.,
PodunavacKuzmanović Sanja O.,
Tepić Horecki Aleksandra N.,
Kovačević Strahinja Z.,
Vidović Senka S.,
Šumić Zdravko M.,
Ilin Žarko M.
Publication year - 2018
Publication title -
journal of food processing and preservation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.511
H-Index - 48
eISSN - 1745-4549
pISSN - 0145-8892
DOI - 10.1111/jfpp.13355
Subject(s) - chemometrics , lactuca , carotenoid , antioxidant , partial least squares regression , ranking (information retrieval) , linear regression , chemistry , multivariate statistics , food science , phenols , predictive modelling , mathematics , botany , machine learning , biology , statistics , chromatography , biochemistry , computer science
The aim of this study was to establish models for the antioxidant activity (IC 50 ) prediction of lettuce samples ( Lactuca sativa L.) using their phytochemicals content (total phenols, chlorophylls, total carotenoids, and vitamin C) by using suitable chemometric methods. In order to select phytochemicals that best describe the antioxidant activity, classification, and regression analysis were conducted. Generalized pair correlation method (GPCM) gave insight which phytochemicals are the most important for the antioxidant activity prediction. That was confirmed using multiple linear regression (MLR) and establishing models that were validated. The sum of ranking differences (SRD) was applied in order to obtain and rank models with best predictive power. Both GPCM and MLR showed that the most important phytochemicals for the antioxidant activity prediction are chlorophyll a, chlorophyll a + b, and total carotenoids content. The obtained results should be considered as preliminary for the prediction of the antioxidant activity and further improvement in scientific research on this subject follows. Practical applications The models for the prediction of the antioxidant activity were obtained. Novel approach for variable selection was conducted and new method for the MLR model ranking was applied. Chemometric guidelines for the selection of lettuce cultivars with increased antioxidant activity were defined.