Sensitivity Analysis in Predictive Models for assessing the Level of β-Glucan in Oats and Barley Cultivars Using Meta-Models

Oats and barley β-glucans are well-known for their many health benefits; this has encouraged the food industry to develop new functional foods containing oats and barley. This study aims to develop an advanced sensitivity analysis to analyse and evaluate the most significant model inputs contributing to uncertainty in assessing the level of β-glucan content in harvested oat and barley grains. Two methodologies, nominal value and regression method sensitivity analysis, were adopted. The nominal sensitivity analysis highlighted that cultivar selection is the predominant factor with a correlation coefficient 0.66 for hulled oats and barley cultivars, whereas the correlation was 0.80 and 0.77 for naked oats and hull-less barley, respectively. Advanced sensitivity analysis using regression modelling highlighted that cultivar selection, storage days and germination time (days) were the most important parameters in both the oats and barley model. Regression analysis using the response surface methodology shows that prediction models were found to be significant (P < 0.0001) with low standard errors and high coefficients of determination (R ² > 0.94). This study shows that regression modelling is an effective tool to highlight the effect of key input variables and their interactive effects on the predictive response of β-glucan in harvested oats and barley cultivars.