Emily BleiweisThree years ago, Debi Hudgens, PhD, MBA, CLP, CA-AM, was diagnosed with a rare renal disease.
Hudgens is fortunate, she said — there are new medications that have come out in recent years that are helping. But since her diagnosis, Hudgens found herself in the rare disease space, something that caused her to realize just how much work still needs to be done for patients struggling.
“There’s a lot of medications that still need to be developed. That’s my dream — to go work in that space and really come up with new therapeutics,” said Hudgens, a chemist by trade and a student at the University of Maryland School of Pharmacy (UMSOP). “Whether it’s in the early discovery [phase] or even in the … translational science [phase], finding new biomarkers [and] doing clinical trial optimization for those rare disease trial patients — any of those are areas where I’d really like to be to be able to help out.”
It’s that goal to help others dealing with a disease with little or no treatment options that led Hudgens to UMSOP’s newly launched, one-of-a-kind MS in Artificial Intelligence for Drug Development (AIDD) program.
The MS in AIDD was born from industry experts like Joga Gobburu, PhD, MBA, professor, Department of Practice, Sciences, and Health Outcomes Research at UMSOP, looking to educate and prepare students with cutting-edge skills in natural language processing and machine learning to accelerate innovation at every stage of drug development.
The program, led by Gobburu, recently completed the fall semester with its first cohort of 18 students — Hudgens included.
‘Like Looking for a Needle in a Haystack’
Hudgens’ story isn’t unique.
One in 10 Americans lives with a rare disease, half of whom are children, according to the National Organization for Rare Disorders. But, of the more than 10,000 known rare diseases, fewer than 5 percent have an approved treatment.
But what if drug developers were better able to sort through the insurmountable amount of data around medications to know what drugs approved for one disease may be usable for other diseases?
“Imagine in drug development: You have [national and] global regulatory agencies that keep putting out documents, regulations, policies, [and] product labels. Easily … you’re looking at millions of documents,” said Gobburu. “Can you, as a drug developer, use AI to … not only access documents, but be able to pinpoint the answers to questions that you’re looking for?”
“It’s like looking for a needle in a haystack. That is where AI can help drug developers,” he added.
That ability to sort through large amounts of data — be it to find drugs that could be used off-label for something else, scan through in-patient medical records to help diagnose patients in hospital settings quickly, or help understand side effects and safety in clinical trials — is where AI can make a big difference.
For example, it would normally take a team of experts a lot of time to prepare dossiers about the market landscape for a new drug, said Adarsh Subbaswamy, PhD, an assistant professor of Practice, Sciences, and Health Outcomes Research at UMSOP and faculty member of the AIDD program.
This includes doing regulatory intelligence about existing drugs, their labels, and what studies they used, said Subbaswamy, who prior to this role worked for the Food and Drug Administration. AI can scan through the research and help come up with trial protocol.
What AI cannot do, Subbaswamy said, is replace human expertise and oversight. That’s where UMSOP’s program comes in.
“You need the human expert here in order to understand: Why does this make sense? You need to be able to justify the information and the choices. You can’t have this entirely hands-off approach of where you just let AI make a bunch of decisions for you and automate this entire process,” he said. “Ultimately, the human professionals here take ownership for the work. And so that’s why when we talk about responsible use, it’s really about understanding what you can’t automate away: expertise and … liability.”
Learning Through a Pharmaceutical Lens
Programs focusing on AI and machine learning aren’t new, but UMSOP’s approach is. Instead of a degree that trains students in programming, using software, and the different methodologies for artificial intelligence, Gobburu said, the AIDD program flips the model.
“You may have data scientists and computer science graduates, who may be good with the AI part, but they don’t have experience in drug development. But then we have a vast workforce who is experienced — not just trained, but experienced hands-on in drug development,” he said.
“So we want to capture both ends of the spectrum where this program would retool the existing folks who have knowledge in drug development and they would like to gain skills in AI so that they can be more contemporary and also be competitive professionally. … The other end of the spectrum, there are, in fact, a few students who are engineers and data scientists, but they don’t have experience in drug development. So, they also come to the program because here we teach them basics of drug development as well as the AI methodology. All courses adhere to our Decisions-Information-Analysis paradigm, which instills systematic approach to making decisions.”
The UMSOP program is really focused on that application aspect, he added.
Joe Fitzgerald, who has spent nearly three decades in the pharmaceutical industry and now serves as head of data at a biotech company, enrolled in the program to deepen his understanding of how AI can be applied strategically across the drug development process.
He recognizes the importance of deeper domain knowledge. This degree highlights the importance of not just having technical fluency but also combining it with strategic skills, he said.
“What I really appreciate about the program is that every course, from the AI methodology side to optimizing drug development portfolio strategy, is rooted in the real-world pharmaceutical pipeline,” Fitzgerald said.
That integration, he noted, is what differentiates graduates and enables them to contribute at a more strategic level. By combining technical AI capabilities with a deep understanding of the drug development lifecycle, the program equips students to contribute to portfolio decision-making, shaping how and which therapies should be developed.
Other courses he’s taken don’t include how to implement what is introduced, he said.
“I think that’s one of the things that’s a real differentiator for this degree program. In the industry, people often learn concepts or terminology in isolation, gaining an awareness of the tools but not how to apply them in practice,” he said. “But if you don’t understand how that knowledge fits into the full lifecycle, from discovery through clinical development, you can’t fully leverage it — in this case, to correctly identify where AI provides meaningful decision-support for real drug development program decisions.”
