Machine learning to quantify techno-functional properties - A case study for gel stiffness with pea ingredients

Quantifying techno-functional properties of ingredients from multiple crops using machine learning

“…The trends of the fitted individual models were revealed using a theoretical dataset, in which one component increases, while the other components remain constant at 2wt% ( Fig. 5 ) ( Lie-Piang et al, 2023 ). From this Figure, it can be derived that protein and fibre in lupine seed have a very different contribution to the foaming capacity, and to a lesser extent for unheated viscosity and emulsions stability.…”

“…The solubility of the protein in each ingredient was expressed with the nitrogen solubility index (NSI%) and was determined using an adopted protocol ( Lie-Piang et al, 2023 ). Each ingredient was mixed thoroughly in a 1 wt% dispersion and rotated for 30 min, after which they were centrifuged (30 min, 10,000 g, at 20 °C).…”

“…The gel stiffness in terms of Young's modulus was measured according to an existing protocol ( Lie-Piang et al, 2023 ). The ingredients blends were initially dispersed for 5–10 min and then hydrated for 1 h while stirring gently and covered with parafilm.…”

“…The relations between the selected ingredient blends and the techno-functional properties were quantified using a range of machine learning methods according to the framework presented in a previous study ( Lie-Piang et al, 2023 ). For a detailed description of each method we refer to this article.…”

“…In this study, formulations with different compositions and environmental impacts could be generated using the functionality-driven formulation approach using multiple linear regression with the composition of the ingredients as independent variables. To quantify other functional properties that are not linear, a framework was proposed to select a suitable machine learning algorithm ( Lie-Piang et al, 2023 ). The models were subjected to the condition that the predictions were physically feasible, such as the absence of artefacts that could impede with the formulation.…”