Machine learning-powered electronic nose classifies food spoilage and allergens with 92.6% accuracy.
An “electronic nose” created by UC Berkeley researchers can detect the gases produced by spoiled food and food allergens better than human noses.
June 17, 2026 | Article Published in UC Berkeley News (linked)
BSAC Co-Director Professor Ali Javey, BSAC researcher Carla Bassil, and collaborators have published a new paper in Science Advances describing a scalable electronic nose that combines a heterogeneous array of gas sensors with machine learning to classify food-related scents. The team demonstrates automated identification of food spoilage and nut allergens using a 16-element sensor chip, achieving an overall prediction accuracy of 92.6%. The work addresses a longstanding challenge in developing scalable, multiplexed gas sensor arrays while advancing new approaches to scent-based sensing.
Beyond the laboratory, the technology points toward practical applications in food safety and consumer sensing. The accompanying UC Berkeley news stories highlight Bassil's portable electronic nose, which can be operated through an iPhone app, as well as longer-term concepts such as smart refrigerators that monitor food freshness and smartphone-based tools for screening food for allergens. The research team is continuing to evaluate the technology in more complex, real-world environments as the platform evolves.
Read a profile of Carla Bassil on the College of Engineering website.