Artificial intelligence algorithms coupled with top data science can hunt cancer faster than an expert radiation oncologist, new research published in JAMA Oncology shows.
The AI-based solutions created to target tumors for radiation therapy were developed using crowd-innovation techniques in the form of an online challenge with cash prizes.
Contestants across the globe submitted algorithms with the hope of improving cancer care by copying the accuracy of an expert oncologist in targeting lung tumors for precise radiation therapy — a process that takes doctors an average of 8 minutes and a computer just seconds.
“We can develop AI techniques to replicate the expert skill set of a human doctor in a specific task in treating cancer,” said Dr. Raymond Mak, radiation oncologist at Brigham and Women’s Hospital/ Dana-Farber Cancer Institute and first author of the research.
Mak said artificial intelligence won’t necessarily replace doctors, but assist them.
“Really what I imagine for the first applications of AI like this is to help support centers that don’t have the same level of expertise,” said Mak. But he added that the technology still needs time to fully develop.
“It’s like the driverless car situation …These machines are trained to replicate my human capabilities, they’re probably not quite ready to fly on their own,” he said.
Mak said that with the help of AI, doctors will spend less time on the computer and more time with patients.
“I didn’t go to medical school just to be on a computer, I want to help people,” said Mak.
A slice-by-slice tumor segmentation performed manually on a CT scan can be extremely time-consuming, and there is currently a global shortage of radiation oncologists who can perform such a task.
“With a partnership with a good AI, maybe the AI can take seconds to do that work, get me 90 percent of the way there, I touch it up a little bit and then that frees me up,” said Mak.
AI solutions can increase productivity for busy clinicians and transfer expert-level knowledge for education and training purposes as well. The technology has also been successfully applied to other areas of medicine like diagnosis of skin cancer from photos or breast cancer using mammograms.
“We’re on the edge of, I think, explosion,” said Mak. “Five years from now I think that much of what I do in radiation oncology and radiation therapy planning will be mostly automated by AI.”
Check out Alexi Cohan’s new “On the Mend” medical podcast every Thursday or anytime on social media.
If you missed On The Mend today, you can check it out here! https://t.co/iaYc17pysq
— Alexi Cohan (@lexcohan) April 18, 2019
