Machine learning can help identify suitable distribution and storage conditions for delicate extra virgin olive oil
Nowadays, in the United States, and especially in California, different extra virgin olive oils are being produced. Nevertheless, many European countries have been and still are importing their extra virgin olive oils, mainly from Spain and Italy. Many recent studies performed in the US regarding the quality of these imported oils have revealed many low-quality products, even from prestigious brands, causing much discomfort in this sector overseas. Although initial blames were put on the producers, there is more and more evidence that, in fact, the distribution chain and storage plays a major role in the degradation of these delicate foods. Harsh conditions have been reported during transportation in ships, having olive oil containers sitting in the sun from days to even months at very high temperatures.
It is known that extreme temperatures lead to a faster degradation of extra virgin olive oils, as the present pigments and antioxidants are highly sensitive to these variations. To find out whether how temperature during storage might have to do with olive oil quality, machine learning experts from various institutions, including San Diego, California’s Scintillon Institute, conducted a study involving two very different Spanish brands: Almazara del Ebro and As Pontis; Empeltre and Manzanilla varietals, respectively.. During the course of three months to monitor the evolution of products’ quality in different temperature conditions, researchers used visible absorption spectroscopy to generate the raw data and they saw that hot and cold temperatures (e.g. 40 °C and 3 °C) led to a faster degradation than when the oils were kept at room temperature (~23 °C). This information was used to “feed the machine”, in another word to train an artificial neural network (ANN), a machine learning-based algorithm, to estimate the time a specific product had been in a particular temperature. As results, the intelligent algorithm learned how to estimate the temperature conditions and exposure duration an olive oil sample had undergone, solely employing data from the light absorption data analyses.
Being able to perfectly determine whether the product had been kept in hot or cold conditions, and with an error of around ± 5 days of exposure, this optimized “machine” has revealed a potential tool to monitor the conditions suffered by the product during their transport or shelf-life. Combining this simple analytical approach with sophisticated algorithms can help reveal the optimal shipping and storage conditions, as well as locate the phases of the distribution chain where they suffer heavy quality losses. This work would enable extra virgin olive oils, and most likely other kinds of foods or products, to reach the clients in a state that truly reflects the producers’ abilities and care.
*To find the full description about this work, the following scientific article should be searched:
Aroca-Santos, R., Lastra-Mejías, M., Cancilla, J. C., & Torrecilla, J. S. (2018). Intelligent modeling to monitor the evolution of quality of extra virgin olive oil in simulated distribution conditions. Biosystems Engineering, 172, 49-56.
Cancilla, J. C. is a postdoc fellow in Professor Jiwu Wang’s lab at Scintillon Institute. His work has been funded by Allele Biotechnology.