STORY BY MICHELE COFFILL / PHOTOS BY KENDRA STANLEY-MILLS

For people who have epilepsy, it’s often a trial and error process through several of the 30-some antiepileptic drugs to get to a medication that works.

Imagine, though, a situation where physicians and health care teams could explore a patient's dashboard — uploaded using artificial intelligence tools — and then prescribe personalized medication without trial and error. What if those same AI tools could help patients with rare diseases?

Zachary DeBruine, assistant professor of computing, and a team of student researchers are exploring how to develop and deliver artificial intelligence tools to help clinicians at Corewell Health and Helen DeVos Children's Hospital who treat patients with rare diseases. 

Dr. Caleb Bupp, who leads Corewell Health's Medical Genetics and Genomics Division, said DeBruine's research could be invaluable in a medical field where 1 in 10 Americans has a rare disease yet 95% of patients have no treatment.

”We're cooking with gas now,” Bupp said.

DeBruine joined Grand Valley’s faculty in 2022, after working for nearly a decade at Van Andel Institute (VAI) as a post-doctoral fellow and adjunct faculty member. DeBruine’s work at VAI was groundbreaking in helping researchers store and analyze massive sets of biomedical data. It also provided a foundation for his work at Grand Valley centered around building generative AI tools for clinicians and researchers.

”We’re trying to develop genomic intelligence for clinical practice,” DeBruine said. ”This will put AI tools that understand not the English language like ChatGTP but the language of genetic regulation and genetic variation in the hands of clinicians so they can ask questions and get answers that help them improve patient care.”

Bupp said one of the many challenges of working with rare disease patients is the lack of comparisons to other patients. With DeBruine’s genomic intelligence tools, Bupp said he and his team are uncovering layers of understanding with patients.

For example, Bupp explained one ultra-rare condition that affects only 16 people worldwide. ”We can treat a few people by repurposing a drug we know works well. With Zach’s team, we're getting beyond the chemical reaction to the ’Why did this work?’” he said.

Bupp said genetics as a field used to be very reactive. With the advance of AI, ”We are shifting to being more proactive and Zach and his team are helping us to do that. They’re not giving us a road map but they're getting us in the driver's seat,” he said. 

The three organizational partners, Grand Valley, Corewell Health and VAI, have provided support and funding to create AI models that can provide collective insight into rare diseases.

Their first focus was pediatric epilepsy, a subject DeBruine understood well. His son, Andrew, was diagnosed with the disorder at birth. DeBruine explained to an audience during a Grand Rapids Tech Week event that Andrew underwent three different drug therapies, which is not uncommon for patients with epilepsy. Andrew is now 8 and healthy and happy, DeBruine said. 

In the AI model DeBruine and his students are building, results from a simple blood draw would be uploaded to an app that generates an AI cell atlas that can be explored through a dashboard for physicians and clinicians to prescribe personalized medication to treat, for example, Andrew’s epilepsy. 

”If we can learn a model that understands the language of genetics, that model would be applicable not only to pediatric epilepsy but also rare disease diagnosis,” DeBruine said.