A research team from the universities of Seville, Cordoba and Turin and the Cabra center of the Andalusian Institute of Agricultural Research and Training, Fisheries, Food and Organic Production (Ifapa) has applied artificial intelligence to predict with high accuracy the aroma characteristics of oil based on its chemical compounds. This model, trained with thousands of data, makes it possible to classify oils by sensory quality without the need for human tastings, which could revolutionize quality control processes in the oil industry.
The method they propose can clearly distinguish between different varieties of olives and also between those harvested at different stages of ripening. Specifically, in the article “Beyond current quality indices: Quantitative volatilomics unrevealed cultivar traits, harvesting practices impact, and aroma blueprint of extra-virgin olive oils” in the Journal of Food Composition and Analysis, the researchers show how different aromatic profiles are established, even when sensory differences are subtle or imperceptible to the taste and smell of tasters.
Using advanced chemical analysis techniques, participants in this research, funded through the project “Differential markers of organic extra virgin olive oils: physicochemical and sensometric characterization of oils from the main varieties of Andalusia” of the Ministry of University, Research and Innovation of the Andalusian Government, have managed to ‘smell’ the oil with a precision that surpasses the human sense itself.
"The smell of the oil not only delights the palate, but also reveals key information about its quality, olive variety, cultivation method and even its place of origin. Thus, each oil has a kind of ‘aromatic fingerprint’ that defines and characterizes it," said University of Seville researcher Raquel María Callejón, author of the article.
One of the most outstanding advances of the study is the use of artificial intelligence to predict the aroma of the oil from its chemical composition. By combining chemical data with sensomics tools, the science that studies how compounds affect our senses, the researchers have developed a model that can anticipate what sensory attributes an oil will have without the need to taste it beforehand.
This not only opens new doors for the olive oil industry, but also offers additional guarantees to consumers who are looking for high quality, authentic products that are aligned with their values, such as respect for the environment.
Aromas and olfactory compounds
In the study, the scientists analyzed more than 190 aromatic compounds, known as volatiloma. To do so, they have taken into account different samples of extra virgin olive oils, particularly extracted from the picual and hojiblanca olive varieties, the most cultivated in Spain, according to data from the Ministry of Agriculture, Fisheries and Food.
The laboratory technique on which they have based their work is called two-dimensional chromatography, used to separate, identify and measure the different chemical compounds in a sample. In other words, it manages to disassemble the odor in each of its components, indicating which ones are present and in what exact quantity they are found.
On the other hand, the researchers applied Principal Component Analysis (PCA) and Partial Least Squares Discriminant Analysis (PLS-DA), mathematical tools used to understand complex data and reduce the number of variables without losing important information. It could be said that with these machine learning techniques a summary is made of the multitude of data obtained by grouping them into similar patterns. In this case, it is a ‘magnifying glass’ that reveals the map of olive oil flavors according to different variables, such as the olive variety, the type of crop, its origin or the degree of ripeness.
Organic farming with fingerprint
The studies showed that the method of tillage, conventional or organic, influences the aroma of the oil. In the case of the picual variety, two different groups were clearly identified according to the type of cultivation by using statistical techniques and machine learning models.
Oils produced with organic farming showed a higher presence of compounds such as (Z)-3-hexenol and 2-pentanol, which are associated with fresh and herbaceous aromas. However, conventional oils had more alkenes and carbonyls, compounds associated with riper odors. “The results determine that the way olives are grown not only affects environmental sustainability, but also the differential markers between organic and traditional cultivation,” says the expert.
Among the many compounds detected, some act as true markers of quality or origin. For example, (Z)-3-hexenyl acetate is linked to the smell of green leaves and is more abundant in organic oils of hojiblanca, which could be related to the activity of certain substances typical of organic farming.
Based on these findings, the experts propose the possibility of validating the methodology in different products and on a larger scale, consolidating its usefulness both in applied research and in production environments.
Sentido Común Internet
Agrifood Comunicación
Solo disponible accediendo desde Safari