Development of Quality Evaluation Model to Predict Sensory Scores from Measured Viscoelastic Characteristics and Aroma Components for Cheddar Cheese

A quality evaluation model for Cheddar cheeses has been proposed to identify interrelationships among the database of viscoelastic parameters and aroma components obtained from instrumental measurements and both sensory evaluation and palatability scores by utilizing the combination of stepwise PLS-VIP method and ANN modeling. Viscoelastic parameters and aroma components were measured by fracture, creep, texture tests and GC-O analysis, respectively. Sensory evaluations by analytical panel and consumer palatability test were carried out by applying the selected sensory terms to evaluate the quality of Cheddar cheese. Regression equations to predict the sensory and palatability scores were determined by applying PLS analysis to the instrumental measurement data, and thus the contribution degree and direction of explanatory variables against objectives were revealed by stepwise PLS-VIP method. Results indicated that sensory and palatability scores were affected by characteristic parameters belonging to different quality attributes (viscoelastic and aromatic), and thus interrelationships among them caused by the existence of mutual effects among the quality parameters were successfully demonstrated. Finally, interrelationships among parameters were determined with high accuracy by the method, which directly applies the quality characteristic parameters selected by stepwise PLS-VIP to ANN modeling. The proposed directly coupled method is expected to be a novel and useful tool that limits the labor and time requirements involved in developing new food products instead of the repeated operations for preproduction sample designs and sensory evaluations.