Valorization of artichoke processing by-products via production of oligosaccharide-enriched extracts optimized by artificial neural networks

Valorisation of agro-industrial by-products is central to circular-economy strategies. This study evaluates the potential of industrial artichoke by-products to yield oligo-fructan (OS-Fr)-enriched extracts and optimises production using artificial neural networks (ANN). Two by-product types were compared: artichoke powder from cooked by-products (A-PO) and artichoke purée from uncooked by-products (A-PU). Soluble-fibre extraction (80 °C, 60 min; solid loads 46 and 92 mg/g DM) and enzymatic hydrolysis with endo-inulinase N960 were applied in two routes: (i) extraction followed by hydrolysis and (ii) direct single-stage enzymatic treatment. Carbohydrate composition was characterised by HPSEC-ELSD and GC-FID. A-PU provided the highest extractable inulin and low-DP fructans, whereas A-PO was richer in xylan and pectic material. ANN optimisation (factors: temperature, time, enzyme dose) yielded an operational optimum at 61 °C, 1.4 h and 376 U/g DM. Under these conditions, the direct route produced OS-Fr (DP 2-5) 116 mg/g DM A-PU, achieving a higher mass recovery (∼79.6%) than the two-stage approach (56.5%). The resulting extracts combined OS-Fr, pectic oligo-, polysaccharides and arabinans, offering multifunctional technological and potential prebiotic properties but requiring consideration of monosaccharide content and reproducibility. The residual solid fraction was enriched in hemicellulose and pectin, suggesting cascade valorisation opportunities. ANN provided superior predictive performance to Response Surface Methodology (RSM). Overall, direct enzymatic valorisation of A-PU is a promising, simplified route to obtain functional oligosaccharide-enriched extracts within integrated biorefinery concepts.