IBM’s Watson is helping doctors identify treatment options for cancer patients — and it’s even suggesting solutions that humans failed to see.
Researchers at the University of North Carolina (UNC) Lineberger Comprehensive Cancer Center had the supercomputer Watson analyze “large volumes of data,” including past studies, databases, and genetic information, in an attempt to identify treatment options for 1,018 patients with “tumors with specific genetic abnormalities.” The researchers then compared Watson’s treatment choices with those made by a board of cancer experts.
The results were surprising. Not only did Watson confirm the 703 cases in which the expert panel identified “actionable genetic alterations,” its cognitive computing discovered “potential therapeutic options” for 323 additional patients. Human doctors had not identified “recognized actionable mutations” in 96 of these patients.
“To be clear, the additional 323 cases of Watson-identified actionable alterations consisted of only eight genes that had not been considered actionable by the molecular tumor board,” the report’s corresponding author, William Kim, MD, an associate professor at UNC’s medical school, said to the university.
Although the doctors hadn’t considered those eight genes, Watson identified clinical trials for most of those patients — including one that began within a week of the computer analysis.
“Our findings, while preliminary, demonstrate that cognitive computing might have a role in identifying more therapeutic options for cancer patients,” Kim said. “I can tell you that as a practicing oncologist, it’s very reassuring for patients to know that we’re able to explore all possible options for them in a very systematic manner.”
The results are promising for the future of precision cancer care, which treats the disease by developing an individualized plan based on each patient’s genetic information.
“To my knowledge, this is the first published examination of the utility of cognitive computing in precision cancer care,” Kim said. “I’m optimistic that as we get more sequencing data, well-annotated treatment information, as well as therapy response, tools like Watson for Genomics will begin to show their true promise. But, of course, we still need to formally answer these questions.”