More than 55 million people worldwide now live with Alzheimer’s and related dementias. Without treatments or cures, that number could rise to a staggering 152 million people by 2050.
Yiyu Shi, professor of computer science and engineering at the University of Notre Dame, is co-leading a team of eight researchers from Notre Dame and Indiana University that has advanced to the third and final round of a global competition seeking to accelerate breakthrough discoveries by using existing Alzheimer’s data.
The competition, known as the Alzheimer’s Insights AI Prize, is being offered by the Alzheimer’s Disease Data Initiative (AD Data Initiative), which was founded by Bill Gates and dedicated to finding new diagnostics, treatments, and cures for Alzheimer’s.
The competition’s specific mandate is the search for innovative agentic AI innovations.
“Agentic AI systems are capable of independent planning, reasoning, and action,” explains Shi. “That makes them uniquely suited to tackle Alzheimer’s research challenges—where vast amounts of data exist across multiple platforms. Similar strategies are being used with cancer and with cardiovascular disorders—also diseases with multiple biological pathways and many different underlying causes in patients.”
All teams in the competition, including the other remaining finalists—MIT/Harvard University, Prima Mente, an AI lab based in London and San Francisco, The University of Pennsylvania, and Stanford University/The Icahn School of Medicine at Mount Sinai—must combine expertise in medicine and AI.
“Indiana University is renowned in the area of drug discovery for ADRD [Alzheimer’s and related dementias],” says Shi. “Our team members there provided background and science for ADRD under the leadership of my co-leads, Travis S. Johnson and Kun Huang. Then, my graduate students, Xueyang Li and Shuqing Wu, built the agentic AI tool based on the Indiana University knowledge database, one of the most comprehensive in the ADRD community.”
The team’s proposal contains two key innovations.
First, says Shi, “we propose an agentic AI platform for ADRD drug discovery that is grounded in existing scientific literature, making its reasoning explainable. Every AI-generated suggestion or hypothesis for a specific drug target is paired with evidence from a curated knowledge base.”
In Alzheimer’s research, a drug target is the specific molecule or pathway that a medicine is built to act on. Some of the newest Alzheimer’s drugs are designed to latch onto and clear away the sticky amyloid-β plaques that accumulate between brain cells, while older medicines aim at a different target: they boost levels of chemical messengers such as acetylcholine so neurons can keep “talking” to each other a bit longer.
“In both cases, Shi says, “scientists must first decide which target to aim at and then spend years testing compounds that can safely nudge that biology in the right direction. Our platform is meant to help researchers find those promising targets much faster, in weeks or even days, and with stronger, traceable evidence.”
Second, the team’s proposal includes a scalable AI model capable of running in low-resource settings, leveraging the efficient AI research from Shi’s lab on edge AI, model compression, and hardware-aware optimization. The platform requires no coding, enabling researchers, especially those in under-resourced countries or communities, to engage in advanced AI-driven drug discovery without extensive technical training.
If the Notre Dame-Indiana University team, called Alithea from the Greek word meaning truth, wins the final round, its innovation will be featured in the AD Data Initiative’s tool AD Workbench, a free, secure, cloud-based research environment that allows scientists around the world to share, access, and analyze data across platforms.
The competition’s prize is $1 million in funding to advance the team’s research and further enhance the capability and usability of the agentic AI system.
In addition to Shi, Huang, Johnson, Li, and Wu, the team includes Jie Zhang, associate professor at the Indiana University School of Medicine; Yijie Wang, associate professor in the Luddy School of Informatics, Computing, and Engineering at Indiana University; Timothy Richardson, research professor at the Indiana University School of Medicine; and Jeffrey Dage, senior research professor of neurology at the Indiana University School of Medicine.
—Mary Hendriksen, Notre Dame Engineering
Hero photo shows from left Kun Huang, Travis Johnson, Xueyang Li, and Yiyu Shi. (Photo courtesy of Yiyu Shi.)