Deep eutectic solvent-based green extraction of Strychnos potatorum seed phenolics: Process optimization via response surface methodology and artificial neural network

Abstract

The current research focused on extraction optimization of bioactive compounds from Strychnos potatorum seeds (SPs) using an eco-friendly glycerol-sodium acetate based deep eutectic solvent (DES). The optimization was accomplished using response surface methodology (RSM) and artificial neural networking (ANN). The inde pendent variables included shaking time (A), temperature (B), and solvent-to-feed ratio (C), and the responses were the extraction yield, total phenolic content (TPC), total flavonoid content (TFC), antioxidant activity (DPPH), and antidiabetic activity (α-amylase inhibitory activity). The SPs extracts obtained under optimal conditions (29 min, 40 ◦C and 30 mL/g of A, B, and C parameters, respectively) had 30.43 mg gallic acid equivalents (GAE)/g of dry weight (DW) TPC, 10.99 mg rutin equivalents (RE)/g DW TFC, 26.16 % antioxidant activity and 46.95 % α-amylase inhibitory activity. For all the outputs, the ANN percentage error was less than the RSM percentage error for the predicted values against the experimentally measured values. The results were further supported by the %AAD (% absolute average deviation) and R2 values obtained from RSM and ANN methods. The %AAD for TPC, TFC, DPPH, and α-amylase inhibitory activity by RSM was 7.31, 4.80, 4.03, and 4.36, while by ANN, it was 1.18, 3.90, 1.99, and 2.97, respectively. It is worth noting that despite no statistical difference between the two predictive models, ANN gave closer results to the experimental values. Correlation among various response types showed that TPC and TFC were strongly correlated. This research highlights the efficiency of glycerol-sodium acetate DES as an extractant.