Medicine may be on the cusp of a technological revolution. Artificial intelligence and machine learning hold the potential for advancing the practice of medicine and improving patient treatment and outcomes.
AI is the ability of computers to emulate human thought, and ML, a component of AI, is the guts behind it, said Monica Coley, PhD, senior business development manager with Amazon Web Services and the keynote speaker at MSSC’s May 2 membership meeting on the Innovation Campus at Wichita State University.
AI feels like it developed suddenly, with the recent release of large language models such as ChatGPT and GPT-4 changing the world’s discourse seemingly overnight and raising questions and concerns over ethics, accuracy and ubiquity.
But Coley pointed out that artificial intelligence, which was coined around 1955, hark back to 1308 when intellectuals of the time developed paper-based mechanical means to create new knowledge from combinations of concepts. Other milestones have made their mark: the first radio-controlled vessel in 1898, the first chess-playing machine in 1914.
“We have an AI revolution and it’s moving pretty fast, but it’s actually been around awhile,” Coley said, noting there are more than 170,000 articles in medical literature about AI, including 28,000 last year.
But what are the advantages of AI and its implications for health care? Coley notes that AI is being well-used in medicine today with staggering potential. The Mayo Clinic, for example, has been leveraging AI in cardiovascular disease treatment with an algorithm that was 93% accurate in predicting left ventricular dysfunction. Hospitals such as Brigham and Women’s in Boston successfully used robots during the COVID-19 pandemic to triage patients, take vitals and display video feeds between patients and doctors.
The next level is generative AI, where large language models, which use natural language processing, are the foundation.
“That is a form of AI that can actually generate new outputs,” Coley said. “You feed these large language models everything you know, and they begin to be able to predict, based off patterns, what’s next or what the response should be.”
But there are challenges – both ethically and practically – with this burgeoning technology, and not even today’s experts know what all the implications will be. At the forefront are unpredictable abilities and emerging behaviors that experts are rushing to identify and mitigate. Challenges range from medical liability and implications to privacy, misuse and misinformation.
“AI patterns … can actually start to generate information or make predictions you weren’t trained on and don’t know how to do,” Coley said. “It’s kind of scary in medicine. It’s kind of scary period.”
Thought leaders in health care will have to contend with issues such as clinical application and adoption, bias with algorithms, ethical considerations and the current lack of regulations and standards for AI/ML.
The recommendation by some tech leaders to pause AI development for six months, and whether that is realistic, made up the bulk of discussion from the audience and the MSSC panel on AI that included Coley; Sam Antonios, MD, chief clinical officer for Ascension Via Christi; Mallik Karamsetty, director of tech services for Pfizer McPherson; and Andy Lin, vice president of strategy and innovation for Mark III Systems, an IT infrastructure and AI company based in Houston.
Overall, the panel was optimistic that the creation of AI tools in health care has promise.
“Obviously the use of AI is to get more personal care to patients,” Lin said. “There are a lot of ways from a system perspective in getting folks in the right place, seeing the right physicians.”
Antonios noted that physicians are already using AI in their practices in various ways, some of which they may not realize. Those applications and capacities will continue to expand.