Kimchi's taste can now be predicted without a single bite, researchers say

South Korean researchers have developed a technology that can predict how sour, salty and spicy kimchi is through chemical analysis alone — no tasting required. The advance is expected to open new possibilities for quality control by replacing reliance on human palates with scientific data.

The World Institute of Kimchi announced Thursday that it has built a model capable of predicting the intensity of five tastes — sourness, saltiness, sweetness, umami and spiciness — by linking kimchi's compositional data with sensory evaluation results.

As kimchi ferments, compounds such as sugars, organic acids and amino acids change, altering its flavor. Even at the same stage of fermentation, taste can vary depending on the ingredients and manufacturing method used.

Traditional kimchi taste assessment has relied heavily on trained evaluators tasting samples directly, making it difficult to evaluate large volumes of product quickly and consistently.

To address this, a research team led by Kim Ho-myung at the World Institute of Kimchi analyzed the relationship between composition and taste across 32 commercially available baechu kimchi products.

The team stored kimchi at 4 degrees Celsius for zero, three and seven weeks, obtaining a total of 96 samples. They measured acidity (pH), titratable acidity and salinity, and quantitatively analyzed 44 components including sugars, organic acids, amino acids and capsaicinoids.

At the same time, a panel of trained sensory evaluators assessed the intensity of sourness, saltiness, sweetness, umami and spiciness in the same kimchi samples.

The team then connected the compositional measurements with the sensory evaluation results through multivariate regression analysis to develop a model that predicts taste intensity as a numerical value.

The sourness prediction model achieved a coefficient of determination (R²) of 94.8 percent. Spiciness came in at 79.0 percent and saltiness at 70.9 percent. These figures indicate how well the model explains the variation in actual sensory evaluation results from the data used to build it.

The research team also conducted additional validation using 28 separate commercially available kimchi products not included in the model's development. The predicted values for sourness, spiciness and saltiness aligned with actual sensory evaluation results.

The technology is expected to help kimchi manufacturers manage taste variation between batches and monitor fermentation status.

If the predictive model is further refined by incorporating compositional data from a wider range of kimchi varieties, it could help establish an objective quality control system spanning production through retail.

"Kimchi's flavor is determined by the complex interaction of multiple components, so interpreting that through objective data is important," Kim said. "We will apply and validate the predictive model across a wider range of kimchi to increase its practical applicability."

The findings were published in Food Research International, an international academic journal in the field of food science.