Software to estimate the EVOOs “best before date”

By Maurizio Servili and Tullia Gallina Toschi


The level and the rate of the EVOO quality deterioration are greatly affected by its chemical composition and storage conditions (packaging, temperature, and light exposure), and have been deeply documented to be extremely variable according to those factors. The best before date (BBD) to consume an EVOO to be indicated on label (meaning the minimum storage period within which virgin olive oils retain their chemical and sensory parameters within those corresponding to the declared category) is up to the producer decision and the only other objective information about the age of the oil that can be written (if 100 % of the contents of the container come from that harvest), is the harvesting year (or month). 

How can producers provide a reliable BBD for their EVOO, on a scientific basis, since at now they have no objective methods to orient themselves? The availability of a tool to determine the BBD and its preventive use would represent the best tactic of defence against loss of consumer confidence and the risk of fraud allegations.

To fill this gap, in the framework of the Oleum project, a software able to predict the BBD of a given EVOO at given storage conditions has been developed and validated.

To reach the final goal, prediction models have been built based on a large analytical database produced by following the evolution of selected analytical parameters in a real-time shelf-life experiment on 20 EVOO samples, chosen according to the D-optimal design using the Most Descriptive Compound (MDC) algorithm, in order to cover the experimental domain with oils characterized by low, medium and high concentration combinations of selected fatty acids, hydrophilic phenols and tocopherols.

Those 20 EVOO samples were fully analytically characterized for their legal quality parameters, fatty acids profile, antioxidants (hydrophilic phenols and tocopherols), volatile composition, DAGs, and OSI, thus bottled in glass UVA grade, placed on shelves for 12 months to light exposure (for 12 h/day; 25 °C) and for 24 months in the darkness (25-30-35 °C). Sampling and analyses have been carried out monthly for the light exposure and every two months for the darkness.

The parameter analysed during real-time shelf-life for the model construction were: acidity, peroxide value, K232, K270, ∆K, hydrophilic phenols, α-tocopherol, volatile compounds and fatty acid composition.

The entire data set obtained from the analytical results was used to build two different predictive models, one for the light exposure and one for the dark conditions of storage.

Those predictive models were built by a multivariate statistical approach and the coefficients to calculate the missing time for reaching the threshold values of selected legal/quality parameters, in order to determine the BBDs, were extrapolated through the two models.

The predictive ability was assessed by internal and external validation, showing a satisfactory accuracy in the prevision of the date where the most important quality and/or legal parameters will be overcome (in terms of K270, the concentration of hydrophilic phenols, threshold of impact volatile compounds for the rancid defect).

The software has been developed with a user-friendly interface, so that the inputs to predict the BBD of an EVOO are: sample name, condition of storage (light/dark and temperature), and a selection of analytical parameters. The output is represented by the time for exceeding the threshold value (in months).  The accuracy of the prediction of the BBD is satisfactory for the purpose to provide a tool for the producer needs and for strengthening consumer confidence.