Skip to content
Available Technology


Predicting Antioxidant Synergism via Artificial Intelligence

Inventors: Dr. Carlos Garcia, Dr. Daniel Whitehead, Lucas De Brito Ayres

Market Overview

Lipid oxidation is a major issue affecting products containing unsaturated fatty acids as ingredients or components. In specific, the formation of low molecular-weight species with diverse functional groups has the potential to impart off-odors and off-flavors. This process, also known as rancidity, and can not only impart an unpleasant taste, but also diminish the nutritional value, shelf-life, and the overall quality of products, which ultimately impacts all segments of the supply chain. The impacted products include, for instance, cosmetics, vegetable oils, seafood, processed meat, animal feed, and other food samples. While current approaches to controlling this process through the addition of antioxidants minimizes the potential organoleptic and toxic effects of these compounds, empirically predicting how these mixtures of antioxidants will behave has traditionally been one of the most challenging tasks, often leading to simple additive (or even antagonistic) instead of the desired synergistic effects. To address this current gap in knowledge, a novel artificial intelligence model was developed based on deep learning architecture to both predict the type of interaction (synergistic, additive, and antagonistic) of known mixtures as well as to unveil new antioxidant combinations.


Impacts on consumer health and economic supply chain for products including cosmetics, vegetable oils, seafood, processed meat, animal feed, and other food samples

Technical Summary:

This invention is the first example of using Artificial Intelligence based on deep learning architecture to predict antioxidant interactions. The algorithm was trained using the SMILES notation for the antioxidants and a combination index to account for the interaction. This AI
algorithm pulls from a propriety database of approximately 1100 entries and has been enhanced with abundant experimental data in order to provide suitable predictions with statistical relevance. The proposed augmentation approach leads to a more representative chemical space during the model training, which addresses common overfitting problems due to the use of relatively small datasets. As this novel strategy enables a broader and more rational predictions related to the antioxidant mixtures behavior, it could be used as an auxiliary tool in benchmark analysis routines.


• First AI model to predict how mixtures of antioxidants will behave and potential additive, antagonistic, or synergistic effects
• Accounts for the complexity and multifaceted nature of antioxidant response
• Broader and more rational predictions compared to the traditional, empirically driven approaches

Download Printable PDF

Technology Overview

State of Development





Dr. Carlos Garcia, Dr. Daniel Whitehead, Lucas De Brito Ayres

For More Info, Contact:

Interested in this technology?
Contact curf@clemson.edu
Please put technology ID in subject line of email.


Get Started
Ready to Get Started?

Contact our team at CURF

Contact CURF