Whether it’s self-driving cars or articles that write themselves (but not this one), artificial intelligence (AI) is a hot topic. Its potential seems limitless, especially considering how it could transform the world of medicine and patient care.
But cautions abound and there’s much to ponder when going down the AI health care road.
“I think AI in health care is both overhyped and underhyped,” says Roy Esaki, M.D., a Honolulu board-certified anesthesiologist and the chief health information officer for The Queen’s Health System. “I think in the short term, it’s overhyped in terms of incorporating into day-to-day clinical practice. We need to proceed conservatively.”
But Dr. Esaki says in the long term, “I think it’s underhyped; like the development of the internet, it has the potential to completely transform much of the work that we do.”
One place Dr. Esaki sees great potential is doctors may be able to spend less time doing paperwork and more time with patients. “Modern health care is often mired in paperwork and sitting in front of a computer. There are many studies that show there’s often more time spent in front of computers than in front of a patient. I think that in the end, AI will reverse that.”
AI is already enabling positive results related to radiology and medical imaging. Ezekiel Emanuel, M.D., bioethicist, senior fellow at the Center for American Progress, and co-director of the Healthcare Transformation Institute at the University of Pennsylvania, says there are other AI advancements that, in the short term, will greatly improve medical care.
“I think drug design is going to be one of the big areas,” he says, as well as helping doctors look at potential drug interactions. “You can look up one drug and find 15 different interactions – it’s mind boggling. But AI can take seconds to predict if there’s a drug combination that would put the patient at high risk of landing in the hospital.”
Another area, says Dr. Emanuel, is predicting which patients need the most attention to help prevent complications.
“It’s going to help highlight patients who need more attention. If you can identify high-risk patients, those are the ones you need to spend the most time attending to.”
There are concerns that could slow the advancement of AI in medicine, however. Dr. Esaki says one concern is its accuracy.
“The need for accuracy compared with other industries is a lot higher,” he says. “A single mistake can be catastrophic. We need to be a lot more guarded in terms of the immediate impact and how we use it today.”
“AI isn’t going to be perfect,” says Dr. Emanuel. “I think we’ve already seen in other areas that AI alone isn’t the right answer. AI may perform better than humans for a while, but the combination of AI and humans is the right way to go.”
Dr. Emanuel says there’s great potential for AI in medicine, but also cautions about making too much of it.
“The important thing is to put it in context. It’s going to augment what doctors and nurses and the rest of the health system can offer. That’s going to be a good thing so that clinicians can focus on the right things.”
Art: Garry Ono