Traceability of sweeteners in soy yogurt using linear discriminant analysis of physicochemical and sensory parameters

Abstract

The present study was aimed to design and develop the consumer-accepted soy yogurt using different sweeteners and investigate the possibility of tracing these sweeteners by implementing a classification algorithm. The effects of three types of sweeteners on the physicochemical properties, sensory properties, and prediction of sweeteners in soymilk-based yogurt were investigated using linear discriminant analysis (LDA). Yogurts were prepared by inoculating lactic acid cultures into soymilk containing 0.3% gelatin, one of the selected sweeteners [sucrose, honey, and sucralose (an artificial sugar)], and strawberry flavor. Freshly prepared soy yogurts were then analyzed for their physicochemical and sensory characteristics. The results revealed that the honey samples had a higher protein content (4.69%–4.71%) compared to that in plain yogurt (3.83%). Soy yogurts containing the sweeteners had a relatively reduced fat content, with the lowest fat content exhibited by the honey samples (1.88%–1.91%). Furthermore, the addition of sweeteners resulted in final products with varying ash content and total soluble solids (TSS). In contrast, the least effect was found on lactic acid and pH. The overall sensory results revealed that the strawberry flavored soy yogurt with honey as sweetener could attain acceptable quality upon processing. LDA with two components based on the partial chemical composition of soy yogurts could identify the sweetener source in unknown soy yogurt samples. The discriminatory ability of the LDA model was further validated using unknown soy yogurt samples to predict the sweeteners, and an accuracy level ranging from 80% to 100% was attained in 1000 test runs. Moreover, the model could distinguish the presence of artificial sweeteners from the other sweeteners in the samples with 100% accuracy.