The “Ascend” supercomputer cluster at the Ohio Supercomputer Center, which is being utilized for Dr. Simeng Zhu’s research projects. Credit: Courtesy of Ohio Supercomputer Center
Six artificial intelligence-based research projects aimed to help predict cancer will start development in January with a team of student researchers.
The projects, through the James Cancer Hospital and Solove Research Institute,will be designed to locate tumors or predict the risk of them recurring for a variety of cancers using AI algorithms.
“AI machine learning is getting very popular, and I do think that’s the future,” said Dr. Simeng Zhu, a radiation oncologist who is leading the team.
With around two students per project, he said each will have a three-step process.
“First, we have to start with data curation and then the second is going to be the actual model development, where we actually program the algorithm,” Zhu said. “Once we get to the second phase, then finally it’s going to be the algorithm and validation phase.”
Zhu prioritized choosing student researchers who were eager to pick up new skills.
“The main thing that I look for is not necessarily what skills that they currently have. I think what’s more important is their willingness to learn,” Zhu said. “Their willingness to learn kind of determines the velocity of how much knowledge that we’ll have later on, which is more important.”
The research projects are staffed by students from different majors, creating a variety of different experiences and skills.
Hari Garish, a second-year in computer science and engineering, works on one of the projects.
“I think what would make me a better machine learner is to have more conceptual knowledge about the specific industry I’m trying to tap into,” Garish said.
Zhu said there are also pre-medical students who want to become physicians.
“It’s good to involve them as well because they will be able to provide the clinical aspect of the projects which is also very important in developing these sort of medical AI algorithms,” Zhu said.
“You can go into really any industry and they’ll have some [AI] data analysis or they want to implement AI,” Garish said. “I get to learn from [Zhu] about medicine and also how to implement AI into medicine.”
There are risks associated with implementing AI into the medical field.
“If an algorithm mistakenly classifies a cat to a dog, what’s the big deal, right?” Zhu said. “But if we use that for clinical prediction, then the stakes are higher. The standards are higher.”
