How machine learning empowers collaboration between computer science, math and medical research
Dr. Helen Chen
Professor, School of Public Health Sciences
Faculty of Health
Bing Ho
PhD Candidate, Cheriton School of Computer Science
Faculty of Mathematics
Drug development is an arduous process that costs billions of dollars and can last for years or even decades. Whether scientists are trying to understand the potential interactions of two drugs or develop new applications for an existing medication, pharmaceutical research features frequent wrong turns and dead ends.
An interdisciplinary team of researchers at the University of Waterloo are using machine learning to dramatically increase the speed of drug development. “We have a lot of existing data across a broad spectrum of medical domains, but it’s extremely complex, and often not as complete or extensive as we would like,” explains Dr. Helen Chen, professor of practice in Public Health Sciences. “It’s like a very shallow ocean.”
Chen teamed up with Bing Hu, a PhD candidate in Computer Science, to build a machine learning model that can analyze and synthesize large amounts of pharmaceutical research data and predict a drug’s properties and interactions. To best represent the effect of drugs on the body, they turned to Dr. Anita Layton, a professor of Applied Mathematics who has received international recognition for her research building mathematical models of the kidneys.
“Often, when we use machine learning to train neural networks, we’re starting from scratch,” Hu says. “But by drawing on the enormous amount of domain specific knowledge coming from biology and medicine, we’re able to build more efficient, more accurate models whose predictions consistently match-up with existing data from the real world.”
The team’s model can predict how a drug might interact with a particular protein target, and how the medication might behave in someone’s body in terms of efficacy and safety.
“Personalized treatment is the next frontier in medicine,” Chen says. “Machine learning research like this is putting that treatment in the hands of everyone.”
The research team’s collaborative work isn’t limited to campus. They’re collaborating with other experts around the world to gather data, build hypotheses and make lab and clinical trials more efficient and effective.
In Ontario, they’re working with medical researchers at the Princess Margaret Cancer Centre to best understand how to use their new tech strategically. They’re also working with researchers in the Advanced Data Science Lab at Yonsei University in South Korea to think about the technology’s potential for global impact.
“AI is powerful and exciting, but we need to focus on using it to build tools that will actually benefit people,” Hu says. “That development needs to be a collaborative process where you work with experts to create the tools they need to make the next world-changing breakthrough.”
“One of the most exciting things about this work is that we’re bringing together perspectives from so many disciplines,” Chen says. “That convergence, combined with the power of AI, makes discovery so much faster. It’s like before we were riding a horse from A to B, and now we’re riding in high-speed trains.”
