Aim: Artificial neural network (ANN) development to find optimal carriers (pea protein-P, maltodextrin-M, and inulin-I) mixture for encapsulation of pumpkin waste bioactive (β-carotene and phenolics).
Methods: Freeze-drying encapsulation and encapsulates characterisation in terms of bioactive contents and encapsulation efficiencies, water activity, hygroscopicity, densities, flowability, cohesiveness, particle size (laser diffraction), solubility, colour (CIELab), morphological (SEM), stability and release properties.
Results: Optimal encapsulates, OE-T (with highest total bioactive contents; P, M, and I of 53.9, 46.1, and 0%w/w) and OE-EE (with highest bioactive encapsulation efficiencies; P, M, and I of 45.5, 32.0, and 22.5%w/w) had particle diameters of 94.561 ± 1.341 µm and 90.206 ± 0.571 µm, the span of 1.777 ± 0.094 and 1.588 ± 0.089, highest release at pH 7.4 of phenolics of 71.03%w/w after 72 h and 66.22%w/w after 48 h, and β-carotene of 43.67%w/w after 8 h and 48.62%w/w after 6 h, respectively.
Conclusion: ANN model for prediction of encapsulates' preparation, showed good anticipation properties (with gained determination coefficients of 1.000).
Keywords: Pumpkin waste; artificial neural network; bioactive compounds; encapsulation; optimisation.