Eleven years ago, we reviewed the Pulitzer Prize-winning book, The Emperor of All Maladies: A Biography of Cancer, by oncologist Dr Siddhartha Mukherjee (© Getty Images). As described on the dust cover, the author “examines cancer with a cellular biologist’s precision, a historian’s perspective, and a biographer’s passion. The result is an astonishingly lucid and eloquent chronicle of a disease humans have lived with – and perished from – for more than five thousand years.” Earlier this week, Mukherjee delivered a keynote lecture entitled “A Peek into the Future of Biomedical Transformation” at the annual meeting of the Radiological Society of North America (RSNA) in Chicago.
In his introduction, Mukherjee said, “To start with, I want to talk about deep learning which means deploying learning algorithms that imitate human learning.” He asked, “Can machines learn like us? Can machines learn medicine?” Mukherjee then referred to an article he’d written in 2017 entitled “AI versus MD” that was published in The New Yorker. The editors had provided the title which he felt should rather have been “AI with MD.” A pertinent question was whether AI could mimic how human intelligence works, and the answer will be important for the future of medicine.
Mukherjee next described a series of clinical examples that involved the detection of cancer using AI algorithms as well as leveraging AI to support cancer research. The areas that he believed held the greatest promise were detecting breast cancer using mammography, identifying pancreatic cancer in high-risk patients, and evaluating the likelihood of recurrence of prior cancer. He also believed that AI algorithms had a pivotal role to play in providing a second opinion.
“You could imagine yourself making a diagnosis while the AI algorithm makes a diagnosis, and then you compare notes,” he said and continued, “And so the algorithm could ask you whether you’re sure of your diagnosis.” Mukherjee stated that the most important question that needs to be asked is, “What is the endpoint?” He argued that the best endpoint was lives saved, although he acknowledged this was a measure that required decades of research to answer.
Mukherjee was optimistic about the future of AI in transforming the practice of medicine, particularly in the next few years. This year’s RSNA meeting has certainly provided numerous examples of the utility of AI algorithms, one of which came from Lunit, a Korean company, which showed that AI models using digital breast tomosynthesis performed better than AI models using digital mammography (seen above right, © Su-A Yang).