Targeting oil and fat rancidity: Scientists use AI tools to tackle food spoilage and increase shelf life

27 Oct 2023 --- Scientists in the US are utilizing an AI model based on deep learning architecture to extend the shelf life of food products that go through rancidity, a process that imparts off-odors and tastes to food when it gets oxidized. They say the findings can be applied across F&B applications which use fats and oils and could be particularly useful in the chips, chocolate and nut categories.

Food can go rancid when it’s exposed to the air for too long — a process called oxidation. Many common ingredients, especially lipids, which are fats and oils, react with oxygen.

The study highlights how oxidation leads to the formation of smaller molecules such as ketones, aldehydes and fatty acids that give rancid foods a characteristic rank, strong and metallic scent. Repeatedly consuming rancid foods can also threaten health.

Antioxidants can “neutralize” many of the processes that cause rancidity and preserve the flavors and nutritional value of food for longer, for which researchers have to carefully choose a specific set of antioxidants and precisely calculate the amount to be added to foods.

Chips are rich in fats, making them prone to rancidity.But researchers are stuck with a few mixtures that provide only some level of protection against rancidity. Here’s where AI comes into play.

Leveraging AI 
The study aims to “predict the type of interaction of antioxidant combinations,” which will allow formulators to add lower quantities to antioxidants to boost the total antioxidant capacity of any formulation.

“AI can help us identify the most efficient antioxidant mixtures, thus preventing the fat samples from getting oxidized or rancid,” Dr. Carlos D. Garcia, one of the authors of the paper and professor at the Department of Chemistry at Clemson University in the US, which conducted the research, tells Food Ingredients First.

“The focus of the project is to investigate antioxidant synergism. These are interactions between two or three types of antioxidants that lead to an effect that is much higher than what you would expect with the same amount of individual antioxidants.” 

“While the use of mixtures of antioxidants has proven to be one of the most effective and frequently adopted methods to prevent fat oxidation in foods, predicting how these mixtures will behave has traditionally been one of the most challenging tasks, as it can lead to synergistic, additive or even antagonistic behavior,” says the study.

The scientists used real lard for their experiments. Dr. Garcia added that predicting antioxidant ratios and interactions that would work best to tackle food rancidity is currently challenging, which the team addressed by training an algorithm.

The findings have been published in the Journal of Agricultural and Food Chemistry.

How AI tackles food spoilage
According to the FDA, rancidity in peanuts — which makes them soft, with a yellow, dark, or oily appearance, and a bitter taste and stale odor — is one of the reasons behind 5% (shelled) and 10% (unshelled) or more kernels being rejected after harvest. This constitutes one of the many examples that make controlling fat oxidation essential.

To teach the AI tool how to look for new combinations of antioxidants, the scientists chose a type of AI capable of working with textual representations, which are written codes describing the chemical structure of each antioxidant.

Once the machine could recognize general chemical patterns, like how certain molecules react with each other, the researchers fine-tuned it by teaching it more advanced chemistry. The team used a database of almost 1,100 mixtures previously described in the research literature.

AI models are “dynamic learners” and can be trained to enhance their predictive abilities.“With that database, we trained our algorithm. This process shows the system many “examples” of the process so the computer can make the connections between inputs (mixtures of molecules) and outputs (type of interaction),” underscores Dr. Garcia.

“To test how much the system is really learning, we used a portion of the database that was never shown to the computer. Here, we show the computer a series of mixtures and ask the system to predict the type of interaction that those mixtures would lead to. By comparing the predictions made by the computer with the real interactions, we can estimate the efficiency of the system,” he explains.

To test their AI model, the researchers performed oxidation experiments using real lard (fat from a pig’s abdomen) and phenolic antioxidants.

“Lard is a ubiquitous sample, which makes the applicability of our system much broader. Ultimately, we expect to be able to apply the outcomes of our research to any sample.”

When enquired on what is lacking in the other methods to prevent rancidity in foods, Lucas de Brito Ayres, the lead author of the study told us that the other alternatives to identify the mixtures (such as those based on laboratory experiments) are unable to provide high output.

"On top of that, testing millions of mixtures of antioxidants, at different ratios would take decades. On the contrary, the computer can do run those simulations very efficiently and evaluate that number of mixtures in a few minutes," he explains.

Lucas de Brito Ayres, lead author of the study, with the research team (Image Credit: Clemson University).In other AI-related advances, the F&B industry has been exploring the technology to ensure food safety, conduct health benefit studies, extract fast and distilled market research and insights for F&B brands and even detect diseases in dairy cows.

What lies ahead?
According to the scientists, AI models are “dynamic learners,” something they can use to sharpen AI’s predictive abilities to help formulators innovate better ways to preserve food.

The project is currently exploring more effective ways to train the AI model and looking for ways to improve its competency further.

Ayres flags that the researchers have filed disclosure with Clemson’s IP Office and are contemplating starting their own company to “make sure this system is able to impact consumers” while also being open to partners to implement their findings.

By Insha Naureen