AI With Rhythm: Brain-Inspired Tech Wins 2025 Invention of the Year Award
What if artificial intelligence could think less like a calculator and more like a brain—quick to adapt, sensitive to change and able to learn without being told how?
A team of University of Maryland researchers have developed a method that can do just that: a new way to train AI to mimic the natural neural rhythms of the human brain. The breakthrough, which the team calls “rhythmic sharing,” has the potential to revolutionize AI systems by enabling them to spot early warning signs in everything from cancer to climate, and it claimed the top honor in the university’s annual Invention of the Year competition.
“This work reimagines what AI can be—learning from the brain to create a truly adaptive and intuitive tool, capable of evolving with remarkable potential for real-world impact,” said Physics Professor Wolfgang Losert.
Alongside awardees in three other categories, including a new social innovation category, Losert and co-inventor Hoony Kang Ph.D. ’24 were announced Tuesday as overall Invention of the Year Award winners as part of Innovate Maryland—an annual celebration of UMD researchers’ creativity in translating cutting-edge research into practical, impactful solutions to address global challenges.
“Tonight’s nominees for Invention of the Year are powerful reminders of what’s possible when imagination meets determination,” said Vice President for Research Gregory F. Ball. “They reflect the incredible spirit of innovation that thrives across our research community—a community united by a shared vision to turn bold ideas into lasting impact.”
In addition to the Invention of the Year Awards, three special recognition awards were presented to honor recipients for their exceptional contributions to UMD’s innovation ecosystem and their lasting impact both in the region and beyond.
Kan Cao, professor of cell biology and molecular genetics, was named Entrepreneur of the Year for launching Bluelene; the skin care line from her company, MBlue Labs, is the first to harness the anti-aging properties of the antioxidant methylene blue. [The Age of Skin: Researcher’s Study of Disease That Fatally Ages Children Leads to a Discovery That Could Help Us All Look Younger]
The 2025 Invention of the Year awardees include:
In Information Sciences/Overall Winner: Brain-inspired Method of Machine Learning
Bridging the gap between neuroscience and artificial intelligence, Losert and Kang have developed a radical new way for machines to learn. Inspired by the human brain, their system mimics the rhythmic communication between neurons and glial cells, allowing machines to detect patterns and shifts in data with striking speed and accuracy. The result is an adaptable, efficient learning model that could one day predict financial crashes, natural disasters or cancer mutations, all with a fraction of the information that today’s AI systems require.
In Life Sciences: RNAnneal: Prediction of RNA Tertiary Structure and Thermodynamics With Generative AI
RNAnneal, developed by Millard and Lee Alexander Professor in Chemical Physics Pratyush Tiwary and biophysics doctoral student Lukas Herron, is a breakthrough in computational biology that could reshape how we design drugs targeting RNA—a fast-emerging frontier in medicine. This two-stage, AI-driven pipeline first generates thousands of potential RNA structures, then applies a novel thermodynamics-based model to identify the most stable configurations. It does so without relying on existing RNA data, setting a new bar with 100% prediction accuracy in benchmark tests. By dramatically improving the speed and precision of RNA-targeted drug development, RNAnneal offers new hope for treating cancer, neurological disorders, genetic conditions, infectious diseases and more.
In Social Innovation: AI and Information Literacy Module
As artificial intelligence becomes a regular part of daily life, an interdisciplinary team of researchers, led by Mona Thompson of the Teaching and Learning Transformation Center and Benjamin Shaw of University Libraries, with researchers Hal Daumé III, Katie Shilton and Daria Yocco, want to make sure students aren’t just using AI—they’re thinking critically about it. Their new online module is a crash course on how to understand, evaluate and responsibly use AI tools, offering everything from bite-sized videos to hands-on exercises. The module aims to build digital citizens who can navigate misinformation, cite AI-generated content ethically and maintain academic integrity.
Read the full list of finalists in every category.
Written by Jordan Dunklee
