Story by Madison Ward
Artificial Intelligence, or AI as it’s more commonly called, is revolutionizing many facets of our daily lives. Now AI is promising to improve our health by aiding physicians.
For the past century, physicians have struggled to maintain the reams of medical records and to analyze the trends within a patient’s test results. This data can provide insights into early disease recognition, improve treatment outcomes, avoid adverse drug interactions, and save money on medical care, but only if someone — or something — is able to actually record, review, and remember all of it. And that’s exactly how AI is reforming healthcare.
Dr. Yara Abdou is a medical oncologist at UNC and the Lineberger Comprehensive Cancer Center. She sees AI as a game-changer in helping her patients get healthy and stay that way.
“I believe AI has the potential to revolutionize medicine,” Abdou said. “With the vast amount of data that we have in healthcare, AI can help us make sense of it all and make better decisions.”
One area where AI could be particularly useful is in predicting health outcomes. By analyzing all of a patient’s data, AI algorithms can identify patterns and risk factors hidden within hundreds of results combined with matching the latest research and treatments, a feat mere mortals simply can’t achieve.
Abdou said one of the most common AI technologies used today is the eHealth app.
“This is a technology that helps patients learn about their disease, keep up with the latest data and keep up with their health and well being,” Abdou said. “In cancer in particular, we are looking at eHealth apps that help remind patients to take their pills, for example, or cancer pills.”
Angela Zhang is a medical assistant at MedFirst in Chapel Hill. She says AI is used in preventive medicine to predict health risks.
“What is your risk for stroke down the line?” Zhang said. “What is your risk for certain types of cancer, given your demographic data and family data?”
Despite all of AI’s apparent benefits, there are also challenges to overcome, according to Dr. Saif Khairat, an associate professor at the UNC School of Nursing and the Carolina Health Informatics program.
“One of the problems is bias,” Khairat said. “We all have our biases. And unintentionally, the developer of an AI platform may introduce their own bias. And so that may create inaccurate decision making or inaccurate assessments.”
This makes ensuring AI algorithms are fair and equitable critical, and requires careful consideration of how data is collected and analyzed, and by whom.
Abdou agrees that bias is a potentially significant issue in healthcare AI development.
“There’s a lot of concern around fairness and accuracy of AI models,” Abdou said. “We need to make sure that these models are trained on diverse and representative datasets, so that they don’t perpetuate biases and disparities that exist in the real world.”
Zhang said AI also brings medical ethics into question.
“Let’s say based on demographic data you gave — lab results you’ve been giving to AI system computing systems — predictive technology comes back to show you have high risk of certain conditions,” Zhang said. “Is that grounds for a prospective insurance company to charge you a higher rate because they know that you’re very likely to need extreme or intensive care at some point?”
Despite these challenges, Khairat believes that AI has the potential to make healthcare more thorough, effective, and ultimately less expensive for insurers and patients.
“The next few years will be dedicated to how we train these AI models to be fair, to be efficient, to be accurate,” Khairat said. “So I think that the future is going to be AI for a little bit.”
Meanwhile, Abdou sees AI as part of a broader trend toward personalized medicine, based on an individual’s health factors.
“We’re moving towards more individualized care, where we’re looking at each patient as an individual and tailoring treatment and prevention strategies to their unique needs and circumstances,” Abdou said. “AI is going to be a big part of that.”
