AI systems target tree pollen behind allergies

Imagine just looking at your fingerprints and trying to separate yourself from the same twins. That’s how difficult it is for scientists to distinguish between the small, powdery pollen grains produced by fir, spruce and pine trees.

However, new artificial intelligence systems developed by researchers at the University of Texas, Arlington, University of Nevada and Virginia Tech can make the task much easier and provide great relief for allergy patients.

“When tree species are the most allergic and release pollen, with more detailed data, urban planners can make smarter decisions wherever they want to plant,” said Behnaz Balmaki, assistant professor of biology research at UT Arlington and co-author of the new study published in Big Data Big Data with Masodoo Rostami in the UTA Science Department of Science.

“This is especially important in high trafficking areas such as schools, hospitals, parks and neighborhoods. Health services can also use this information to improve allergy alerts, public health messages and treatment recommendations during peak pollen seasons.”

Pollen analysis is a powerful way to reconstruct historical ecosystems. Conserved pollen grains in the lake bottom and peat marsh provide a detailed record of past plant communities. Plant distribution is closely related to environmental factors such as temperature, rainfall, and humidity, so identifying the types of pollen present in different layers of sediment can help clarify how ecosystems respond to natural climate change over time and how they will respond in the future.

“Even with high resolution microscopes, the difference in pollen is very subtle,” Dr. Barmaki said. “Our research shows that deep learning tools can significantly improve the speed and accuracy of pollen classification. This opens the door to large-scale environmental surveillance and opens up more detailed reconstruction of ecological changes. We also promise to improve allergen tracking by identifying exactly which species are releasing pollen.”

Barmaki adds that the study could also benefit agriculture.

“Pollen is a powerful indicator of ecosystem health,” she said. “Changes in pollen composition can indicate changes in vegetation, moisture levels, and even past fire activity. Farmers can use this information to track long-term environmental trends affecting crop viability, soil conditions, or local climate patterns.

“It also helps to conserve wildlife and pollinators. Many animals, including insects such as bees and butterflies, rely on specific plants to food and habitat. By identifying whether plant species exist or are declining in the region, we can take steps to affect the entire food web and protect the important relationship between plants and pollinators.”

In this study, the team examined historical samples of FIR, spruce and pine trees preserved by the National Museum of History at the University of Nevada. They test these samples using nine different AI models, demonstrating the powerful potential of a technology to identify pollen with impressive speed and accuracy.

“This shows that deep learning can successfully support and even surpass traditional identification methods in both speed and accuracy,” Barmaki said. “But it also confirms how important human expertise is still. We need to have a strong understanding of well-prepared samples and ecological contexts. This is not just a machine, but a collaboration between technology and science.”

For future projects, Barmaki and her collaborators plan to expand their research to include a wider range of plant species. Their goal is to develop a comprehensive pollen identification system that can be applied to various parts of the United States to better understand how plant communities change in response to extreme weather events.