A comparative study on process optimization of betalain pigment extraction from Beta vulgaris subsp. vulgaris: RSM, ANN, and hybrid RSM-GA methods

Beta vulgaris subsp. vulgaris (red Swiss chard) leaf stalks offer a rich source of betalains, natural pigments with promising applications in the food industry. This study employed response surface methodology (RSM) in conjunction with a 3-level Box-Behnken design to optimize independent extraction variables, including temperature, extraction time, and solid-to-liquid ratio, for maximizing betalain extraction from red Swiss chard. Betacyanins and betaxanthins, the key natural pigments, were targeted as response variables. Statistical analysis revealed the optimal conditions for extraction: 21.14 min of extraction time, 52.98°C temperature, and a solid-to-liquid ratio of 21.61 mg/mL, resulting in the maximum extraction of betacyanins (15.53 mg/100g) and betaxanthins (9.5 mg/100g). To enhance prediction accuracy, an artificial neural network (ANN) model was employed, outperforming RSM predictions. Moreover, incorporating a genetic algorithm (GA) into the RSM regression equation predicted even higher betalain contents, with betacyanins reaching 16.53 mg/100g and betaxanthins 10.52 mg/100g. Confirmation experiments conducted under RSM-GA predicted optimum conditions demonstrated mean betacyanin and betaxanthin contents of 16.54 mg/100g and 10.49 mg/100g, respectively. The superior predictive capabilities of the ANN model and the synergistic integration of GA with RSM highlight innovative approaches for enhancing extraction efficiency. Furthermore, the characterized extract exhibit attributes such as aggregated morphology, amorphous nature, and high thermal stability.