Optimization of kidney bean antioxidants using RSM & ANN and characterization of antioxidant profile by UPLC-QTOF-MS

TY - JOUR

T1 - Optimization of kidney bean antioxidants using RSM & ANN and characterization of antioxidant profile by UPLC-QTOF-MS

AU - Yang, Qiong Qiong

AU - Gan, Ren You

AU - Zhang, Dan

AU - Ge, Ying Ying

AU - Cheng, Li Zeng

AU - Corke, Harold

N1 - Publisher Copyright: © 2019

PY - 2019/11

Y1 - 2019/11

N2 - In this study, response surface methodology (RSM) and two artificial neural networks (ANN), multi-layer perceptron (MLP) and radial basis function (RBF), were used for the first time to model and optimize the extraction conditions of phenolic compounds in kidney beans in order to compare and establish effective prediction models. A total of 40 experiments were performed to screen for variables. The highest amount of polyphenols (917.2 mg GAE/100 g DW) and the strongest antioxidant activity (56.03 μmol Trolox/g DW) were obtained under optimal extraction conditions (70 min extraction time, 53% acetone, and 37 mL/g solvent-to-solid ratio). Response values were close to the predicted values with prediction accuracy as RBF > MLP > RSM, indicating that ANN model has higher prediction accuracy than RSM. Furthermore, UPLC-QTOF-MS analysis of extracts under the optimal extraction conditions showed that there were 25 antioxidant compounds detected, 13 of which were proposed from kidney beans for the first time.

AB - In this study, response surface methodology (RSM) and two artificial neural networks (ANN), multi-layer perceptron (MLP) and radial basis function (RBF), were used for the first time to model and optimize the extraction conditions of phenolic compounds in kidney beans in order to compare and establish effective prediction models. A total of 40 experiments were performed to screen for variables. The highest amount of polyphenols (917.2 mg GAE/100 g DW) and the strongest antioxidant activity (56.03 μmol Trolox/g DW) were obtained under optimal extraction conditions (70 min extraction time, 53% acetone, and 37 mL/g solvent-to-solid ratio). Response values were close to the predicted values with prediction accuracy as RBF > MLP > RSM, indicating that ANN model has higher prediction accuracy than RSM. Furthermore, UPLC-QTOF-MS analysis of extracts under the optimal extraction conditions showed that there were 25 antioxidant compounds detected, 13 of which were proposed from kidney beans for the first time.

KW - Artificial neural network

KW - Mathematical model

KW - Phaseolus vulgaris

KW - Response surface methodology

KW - Total polyphenol content

UR - http://www.scopus.com/inward/record.url?scp=85068528053&partnerID=8YFLogxK

U2 - 10.1016/j.lwt.2019.108321

DO - 10.1016/j.lwt.2019.108321

M3 - 文章

AN - SCOPUS:85068528053

SN - 0023-6438

VL - 114

JO - LWT

JF - LWT

M1 - 108321

ER -