Some view artificial intelligence as a threat to job security and quality care, but Stuart Schnitt, MD, pathologist and chief of breast oncologic pathology at Boston's Brigham and Women's Hospital, sees AI as the tool that can lead to better patient outcomes and democratization of care.
AI has the ability to improve workflow and reduce turnaround time by assisting pathologists in diagnosing slides. It can also take over mundane tasks.
"Quite frankly, some of what we do is pretty mundane and because of that, we can sometimes lose our concentration, even though it's imperative that we get things right," Dr. Schnitt told Becker's. "AI can help relieve us from the sort of mundane repetitive tasks and do the things we do but without complaining about how much work it is or how difficult it is."
AI can also act as a quality assurance function that flags slides where pathologists may have missed something. "There may be a lot of false positives, but at least you know, you won't miss anything," he said.
More importantly, it can "level the playing field" by giving confidence to less experienced pathologists and by allowing small hospitals to have the consistency and accuracy of large academic institutions. Trained algorithms, when taught by experts, can provide more consistency in diagnosis, shorten turnaround time and reduce costs by reducing the need for additional stains.
Some pathologists and radiologists are worried AI will take over their jobs, but Dr. Schnitt said "the combination of AI and physician is better than either alone." However, he did say people should always have the last word when it comes to decisions.
"Whenever you have a new technology, everybody thinks, 'Wow, this is great. It's going to take over everything. We're never going to have to think again. This is going to solve all our problems.' But once you start using it, you realize that there are limitations," he said. "We have to use common sense and recognize that these algorithms are adjuncts. It's very useful, but it cannot be used in isolation because that's where you run into danger. That's where we come in as physicians — we still need to maintain control, take all the information from our sources and integrate them into a final determination. Depending on any one thing alone in isolation can lead to a disaster."
Although Dr. Schnitt is excited about the advancements AI can bring, he warned that developing algorithms must be done with care. He said experts in the field must be the ones to establish ground truth that will be used to train AI. "The algorithm is only going to be as good as the people that train it," he said.