UIC Professor Improving Cancer Survival with Artificial Intelligence: An Interview with Dr. Liz Marai
Title Heading link
Artificial intelligence is becoming more of a reality with each new technological advancement. From self-driving cars to disease mapping, artificial intelligence’s numerous applications have only been increasing, and a new project at the University of Illinois’s Cancer Research Center has incorporated AI for cancer treatment. Recently, I had the opportunity to interview one of the lead computer scientists on the team, Dr. Liz Marai, a UIC computer science professor, member of the University of Illinois Cancer Center, and Honors College Faculty Fellow, about her career path and this new technology. Dr. Liz Marai recently received a $2.8 million research grant to develop AI to help prescribe treatment for cancer.
Dr. Marai began with a rich background in 3-D visualization and credits a computer graphics course as an undergraduate majoring in Computer Science, where [she] covered 3D models and vision for piquing her interest and building her experience in 3-D visualization. She found graphics to be “a highly visual way to see and make practical math and physics.”
A memorable example she uses when describing the visualization of math and physics is studying animation errors from Pixar: “Even the bugs are fun—look up “brave bloopers,” “Pixar gag reel,” or “Disney Frozen bloopers” on Youtube to see all those lovely princesses with their hair sticking straight up and so on.” Animation bugs represent the importance of math and physics in visual representations of information.
As Dr. Marai progressed further in the field of computational visualization, she eventually found the chance to bridge the field with oncology research. Clinicians and computer scientists came together in a project run by the National Science Foundation (NSF).
The NSF hoped to pair the two different groups for newer, inventive solutions and ideas to approach “big data in healthcare,” and Dr. Marai recalls meeting and beginning the project with a wonderful group of physicians at the MC Cancer Center in Texas.
Dr. Marai’s current AI project functions similarly to the way an oncologist would when selecting an appropriate treatment plan. “In real life, an oncologist draws on their prior experience with similar patients they have treated in the past, and sometimes on their colleagues’ experience,” notes Dr. Marai.
AI utilizes a similar approach, but instead of depending on human memory, the AI has a large store of information on previous treatments and patients. Using this large store of information, the AI “makes a recommendation for the treatment sequence for a new patient, for example, whether they should go through surgery first and then through radiotherapy.”
Decisions such as determining when a patient goes through radiotherapy and undergoes surgery are a serious matter; these decisions could drastically affect a patient’s quality of life and survival rate, and the AI system improves the process by providing a greater store of data and proven examples to make the best decisions. Oncologists will use the system by consulting it throughout the process.
Dr. Marai has also incorporated her own specialty, computational visualization, into the system. Computational visualization is part of multiple components in the system. Presenting similar patients in previous situations and determining the likeness between their circumstances depended on the system’s computational visualization because the system visualizes the exact location of the tumor in a patient’s body. By gauging how similarly-positioned a tumor was in a previous patient compared to the tumor’s position in the current patient, the system will be able to determine the treatment plans with the most potential for survival.
The system needs “to define what “similar” means in this 3D space and then needs to judge the goodness of that similarity,” says Dr. Marai. “That is a complicated computational process which we were able to automate, and in order to do so we had to use graphics and 3D modeling.” The AI system’s ability to work as an accurate consulting tool for prescribing treatment plans depends on the AI’s ability to visualize the location of the tumors in the body, which is precisely where Dr. Marai’s unique skill set in computation visualization comes in.
“I can see how it would be extremely valuable as a checkpoint for the standard of care, and then even more valuable in clinical settings where the clinician does not have a large oncological team from whose experience they can draw,” she said. This AI system could mark the beginning of a new expected quality of healthcare. Artificial Intelligence is becoming an important part of our lives, and this new AI system is on the cusp of revolutionizing cancer survival.