Response surface methodology and artificial neural network approach for the optimization of ultrasound-assisted extraction of polyphenols from garlic.

This paper aimed to establish the optimal conditions for ultrasound-assisted extraction of polyphenols from domestic garlic (Allium sativum L.) using response surface methodology (RSM) and artificial neural network (ANN) approach. A 4-factor-3-level central composite design was used to optimize ultrasound-assisted extraction (UAE) to obtain a maximum yield of target responses. Maximum values of the two output parameters: 19.498 mg GAE/g fresh weight of sample total phenolic content and 1.422 mg RUT/g fresh weight of sample total flavonoid content were obtained under optimum extraction conditions: 13.50 min X1, 59.00 °C X2, 71.00% X3 and 20.00 mL/g X4. Root mean square error for training, validation, and testing were 0.0209, 3.6819 and 1.8341, respectively. The correlation coefficient between experimentally obtained total phenolic content and total flavonoid content and values predicted by ANN were 0.9998 for training, 0.9733 for validation, and 0.9821 for testing, indicating the good predictive ability of the model. The ANN model had a higher prediction efficiency than the RSM model. Hence, RSM can demonstrate the interaction effects of basic inherent UAE parameters on target responses, whereas ANN can reliably model the UAE process with better predictive and estimation capabilities.