A Hybrid RSM-ANN-GA Approach on Optimization of Ultrasound-Assisted Extraction Conditions for Bioactive Component-Rich Stevia rebaudiana (Bertoni) Leaves Extract

Abstract

Stevia rebaudiana (Bertoni) leaves consist of dietetically important diterpene steviol glycosides (SGs): stevioside (ST) and rebaudioside-A (Reb-A). ST and Reb-A are key sweetening compounds exhibiting a sweetening potential of 100 to 300 times more intense than that of table sucrose. Ultrasound-assisted extraction (UAE) of SGs was optimized by effective process optimization techniques, such as response surface methodology (RSM) and artificial neural network (ANN) modeling coupled with genetic algorithm (GA) as a function of ethanol concentration (X₁: 0-100%), sonication time (X₂: 10-54 min), and leaf-solvent ratio (X₃: 0.148-0.313 g·mL^-1). The maximum target responses were obtained at optimum UAE conditions of 75% (X₁), 43 min (X₂), and 0.28 g·mL^-1 (X₃). ANN-GA as a potential alternative indicated superiority to RSM. UAE as a green technology proved superior to conventional maceration extraction (CME) with reduced resource consumption. Moreover, UAE resulted in a higher total extract yield (TEY) and SGs including Reb-A and ST yields as compared to those that were obtained by CME with a marked reduction in resource consumption and CO₂ emission. The findings of the present study evidenced the significance of UAE as an ecofriendly extraction method for extracting SGs, and UAE scale-up could be employed for effectiveness on an industrial scale. These findings evidenced that the UAE is a high-efficiency extraction method with an improved statistical approach.

Keywords: Stevia rebaudiana; optimization; rebaudioside-A; stevioside; ultrasound-assisted extraction.

Figure 1

Effect of X₁ (ethanol concentration) on TEY (%) target response, (a) the effect of X₁ (ethanol concentration) on ST and Reb-A yields (mg/g), (b) ST yield: sonication time effect on TEY (%) target response, (c) sonication time effect on ST and Reb-A yields (mg/g), (d) Leaf-to-solvent ratio effect on the TEY (%) response, (e) Leaf-to-solvent ratio effect on ST and Reb-A yields (mg/g) responses (f).

Figure 2

Optimal architecture of the multiplayer perceptron (MLP) topology of the developed ANN model. (A) Depiction of the network training curves demonstrating the number of Epochs for the trained subsets for TEY (%) target response, (B) ST yield (mg/g), (C) and Reb-A yield (mg/g) target responses (D).

Figure 3

3D response surface curve and corresponding contour plots of TEY showing the interaction effect of the concentration of ethanol and sonication time at fixed level of leaf–solvent ratio (0.23 g·mL⁻¹). (A) Concentration of ethanol and leaf–solvent ratio at a fixed level of sonication time (32 min) (B) and sonication time and leaf–solvent ratio at a fixed level of concentration of ethanol (50%). (C) ST yield showing the interaction effect of concentration of ethanol and sonication time at fixed level of leaf–solvent ratio (0.23 g·mL⁻¹). (D) Concentration of ethanol and leaf–solvent ratio at fixed level of sonication time (32 min) (E) and sonication time and leaf–solvent ratio at fixed level of concentration of ethanol (50%) (F); Reb-A yield showing the interaction effect of concentration of ethanol and sonication time at fixed level of leaf–solvent ratio (0.23 g·mL⁻¹). (G) Concentration of ethanol and leaf–solvent ratio at fixed level of sonication time (32 min) (H) and sonication time and leaf–solvent ratio at fixed level of concentration of ethanol (50%) (I).

Figure 3

3D response surface curve and corresponding contour plots of TEY showing the interaction effect of the concentration of ethanol and sonication time at fixed level of leaf–solvent ratio (0.23 g·mL⁻¹). (A) Concentration of ethanol and leaf–solvent ratio at a fixed level of sonication time (32 min) (B) and sonication time and leaf–solvent ratio at a fixed level of concentration of ethanol (50%). (C) ST yield showing the interaction effect of concentration of ethanol and sonication time at fixed level of leaf–solvent ratio (0.23 g·mL⁻¹). (D) Concentration of ethanol and leaf–solvent ratio at fixed level of sonication time (32 min) (E) and sonication time and leaf–solvent ratio at fixed level of concentration of ethanol (50%) (F); Reb-A yield showing the interaction effect of concentration of ethanol and sonication time at fixed level of leaf–solvent ratio (0.23 g·mL⁻¹). (G) Concentration of ethanol and leaf–solvent ratio at fixed level of sonication time (32 min) (H) and sonication time and leaf–solvent ratio at fixed level of concentration of ethanol (50%) (I).

Figure 4

Standard HPLC chromatograms of quantified glycoside compounds including ST and Reb-A (a) and UAE extract (b) at optimized process variables of 75% ethanol concentration, 43 min sonication time, and 0.28 g·mL⁻¹ leaf-to-solvent ratio.

Figure 5

Comparison between the experimental values and model–predicted data values that were rendered by the RSM model for: TEY (%) (A) ST yield (mg/g) (B) and Reb-A yield (mg/g) (C); ANN model for: TEY (%) (D), ST yield (mg/g) (E), and Reb-A yield (mg/g) (F), Scatter plot showing the distribution along the straight line of predicted values versus the experimental values that were obtained by ANN and RSM models for the prediction of: TEY (%) (G), ST yield (mg/g) (H), and Reb-A yield (mg/g) (I).

Figure 6

PCA score and loading plots by different target responses (A,B) and the extraction conditions (C,D). The circled clusters showed successful discrimination in terms of target responses and number of experimental runs. Different color in the PCA score plots clusters exhibited peculiarity for each target response analyzed.

Figure 7

UAE comparison at optimized extraction conditions (X₁: 75% ethanol concentration, X₂: 43 min sonication time and X₃: 0.28 g·mL⁻¹ leaf-to-solvent ratio) and CME for: TEY (%) (A), ST, and Reb-A yields (mg/g) (B).

Figure 8

Efficiency comparison on the energy consumption (a) and CO₂ emission (b) from the UAE and CME extraction methods.