AI: The Future of Health Care

Artificial intelligence (AI) – the simulation of human intelligence by machines – is in the technological, political and media spotlights and, depending on who you listen to, could either be the greatest invention since sliced bread or the demise of civilization.

As director of the Center for Artificial Intelligence Research, Metin N. Gurcan, PhD, Professor, General Internal Medicine, spends a lot of time sifting through the hype and hyperbole to find the best ways AI can improve health care. AI research, he says, has been going on for a long time; however, advances in computational power combined with the vast amounts of data available from the internet have helped create very powerful algorithms that are now pretty good at mimicking human intelligence.

“When I start my day, I say ‘I'm here for our patients; what can I do today that will make a difference for our patients?’” Gurcan says. “With the interest in artificial intelligence comes a lot of questions. And people have a lot of hope and expectation for artificial intelligence, but at the same time, they have a lot of concerns.”

Advice and Guidance

The Center for Artificial Intelligence Research officially started in October 2023, and Gurcan sees it as an adviser to Wake Forest University School of Medicine and the Advocate Health organization. He likens the current state of AI to the internet landscape of the late 1990s and thinks how we use it will evolve over time in a similar manner.

“We will realize that some applications of AI are truly beneficial while others are merely adequate,” he says. “It is important to distinguish between the two and focus on maximizing the benefits of AI. As we develop new technologies, educate our researchers and trainees and conduct innovative research, our center can serve as a trusted partner, providing guidance on the correct path forward."

This guidance will be vital, both for keeping up with the rapid evolution of AI and for keeping track of all the ways it touches health systems processes and patient lives. When someone thinks of AI in health care, they typically imagine a provider using it for charting or reading diagnostic images. But the scope is much, much wider.

AI can help schedule appointments, answer questions in a website chat, check symptoms, check for prescription interactions in the doctor’s office and the pharmacy, help process insurance claims, set reminders for follow-up … the list goes on and on. And that’s not even factoring in the reams of data potentially collected by wearable devices from smart watches to continuous glucose monitors to heart monitors.

In all these cases, Gurcan says, people want to know how to interact with AI, make the best use of it, incorporate it into current workflows and make people’s jobs better. To answer those questions, he wants the center to be a resource that helps people figure out the right experts to connect with.

Elevating Healing and Hope for All

Medicine and health care have many effective treatments and strategies, Gurcan explains, but identifying the patients for whom these treatments will work the best can sometimes be challenging. Or maybe a treatment works but very few could afford it. AI can help health care providers decide on the best treatment plan for a specific individual.

“Our center is approaching AI pretty holistically and in a very comprehensive way, because for us, health equity is really important,” Gurcan says. “And we think we can really make a difference in that area using artificial intelligence.

“One project we have with Duke University and UNC-Chapel Hill is developing machine learning methods to identify maternal mortality risks, especially in African American women,” says Gurcan. “How can we identify which women would be at risk, and then how can we develop strategies around helping those women?”

“This is an example of how we can make a difference in health equity, because we know that a certain segment of the population is negatively impacted by the current health system,” he says. “Something is not working, so how can we make use of the power of artificial intelligence to help those women?”

Currently in its second year, the multi-institutional project involves OB-GYN doctors, machine learning experts and data experts. The goals include identifying which patients are at risk, what those risk factors are, what interventions would be effective and when to start those interventions. Gurcan also hopes to use AI to find out the most impactful way to communicate this information to patients and their doctors.

Another current research project involves determining the risk of breast cancer recurrence. One solution is a genetic test that costs $5,000 or more, putting it out of reach for billions of people. The National Institutes of Health-supported research is seeing if AI can analyze hematoxylin and eosin staining, a common pathology slide routinely generated in cancer care, to get a similar set of information as the genetic test. If it succeeds, important information about cancer recurrence risk would become more readily available all over the world.

Gurcan calls human intelligence and its ability to find patterns in data “just amazing.” But when faced with an immense amount of data – for example, 10,000 patient records with geographic information, blood lab results, weight and more – human brains aren’t good at establishing those patterns anymore. AI is. And it can process massive amounts of information exponentially faster than a human can.

More Than Research

Although “research” is right in the center’s name, Gurcan says education and innovation are also important priorities. Medical school students are especially interested in AI, so the center is investigating how to best integrate it into the curriculum without replacing current subject matter. Meanwhile, some medical students are doing AI-related research as part of their Medical Student Research Program, and the center has a summer internship program supported by the National Library of Medicine as well. Last year, many of the 22 interns were women and/or underrepresented minorities.

To raise awareness of AI, the center has planned monthly seminars, called Artificial Intelligence in Action (AI-IA), with different experts discussing various aspects of AI. Topics have included validation systems for AI and machine learning, using AI for health equity, and a seminar with Warren Kibbe, chief data scientist for the Duke Cancer Institute.

And amidst the research and education, the center hopes to collaborate with industry partners. It typically takes many years for a solution to be integrated into clinical workflows, and Gurcan hopes the center can help accelerate that process through collaborations. Working with businesses and industries adds more resources and expertise to innovate solutions and foster implementation.

Keeping a Finger on the Pulse

Anyone in the Advocate Health organization who is interested in learning more about the center and its activities can become a member. Gurcan says you don’t have to be a researcher or practicing clinician to join; just have an interest in AI.

To keep its more than 191 members informed, the center is launching a quarterly newsletter that will include faculty highlights, current research, a publications list and a list of recent grants awarded to faculty. Information about upcoming activities and how to participate will be communicated as well.

“We want people to read those and say, ‘I really liked this idea. I know that they did this in cardiology (for example), but I want to do this in neurology,’” Gurcan says. Likewise, by seeing new grants, faculty – especially early career faculty – with similar ideas might be encouraged to apply.

Gurcan wants anyone who is interested in AI to stay informed and be part of the conversation. Pew Research Center did a study on people’s perceptions of AI. The results, he says, underscore the need for education so that people have a more realistic view that will fuel excitement about good AI and lessen alarm.

“We may develop some really amazing, helpful technologies, but if our patients do not accept those solutions, that’s going to be really challenging,” he says. “We want to make sure that if there is good intelligence and technology, we can really leverage it positively for our patients.”