Optimisation of ultrasonic-assisted enzyme extraction to analyse total flavonoids and antioxidant activity of purple potato using response surface and artificial neural networks model

The present work utilised purple potatoes as the raw material to perform response surface methodology (RSM) and an artificial neural network (ANN) model. The objectives of the present work were to enhance the efficiency of ultrasound-assisted enzymatic extraction of total flavonoids from purple potatoes, and evaluate their antioxidant activity. The results demonstrated that the ANN model achieved a higher predictive accuracy, with a correlation coefficient of 0.99553 than the RSM model (R2 = 0.9919). The optimal extraction process conditions were the addition of 51.34 U/mL enzyme, extraction duration of 36.21 min, and extraction temperature of 53.12°C. The total flavonoid yield was 9.81 mg/g under these conditions, suggesting higher prediction ability of ANN. The scavenging rates of OH and DPPH(2,2-diphenyl-1-picrylhydrazyl) were 81.6 and 61.8%, respectively, for the purple potato extract concentration of 0.24 mg/mL. The present work proposes a novel approach integrating ANN with ultrasonic-assisted enzymatic extraction to predict and optimise flavonoid yields, demonstrating superior accuracy over traditional methods. The findings advance the extraction of bioactive compounds, and highlight ANN's potential for modelling complex non-linear relationships in food science.