Artificial intelligence can already detect strokes, identify patients at risk of sepsis, help doctors document appointments and even serve as a second set of eyes during colonoscopies.
Healthcare leaders say those tools may ultimately give doctors more time to spend with patients by reducing paperwork and other administrative burdens.
Although AI feels like a new and controversial technology, its roots in medicine stretch back decades.
Its origins can be traced to the 1956 Summer Dartmouth Conference on Artificial Intelligence, where data scientists, engineers, and mathematicians shared ideas for AI’s practical applications, according to the National Library of Medicine.
Eliza, the first chatterbot, was developed in 1964, while INTERNIST-1, the first artificial medical consultant, followed in 1971.
Today, those early concepts have evolved into dozens of AI tools already being deployed in Illinois hospitals. For instance, Silver Cross Hospital in New Lenox uses an AI software that rapidly detects clots in stroke patients.
Hospitals across Illinois are increasingly adopting AI tools. At OSF HealthCare, an AI platform quickly and accurately diagnoses diabetic retinopathy, another helps detect sepsis risk, and still another identifies patients likely to benefit from end-of-life care discussions.
But because AI is still a new technology, most doctors are deploying it “very carefully and very thoughtfully,” said Karen Harris, senior vice president and general counsel for the Illinois Health and Hospital Association.
“We still have to remember this is a tool and the humans still need to do a review, at least at this stage,” Harris said. “Twenty years from now, 30 years from now, AI may develop into a different thing. But at this stage, it’s a tool.”
It’s certainly possible AI “may never be the right tool for all situations,” Harris said. But it’s working well in reducing “administrative burden,” she added.
Dr. Nirav Shah, associate chief medical informatics Officer, AI and innovation and director of investigational innovation at Endeavor Health, said Endeavor Health currently has 80-plus active AI tools in its inventory with another “30 to 40 in our pipeline.”
Because “healthcare is extremely regulated,” providers often spend more time on processes than caring for patients, Shah said.
And with the current national shortage of healthcare providers, that situation “is going to get worse,” decreasing both access and safety, Shah said.
Providers typically spend 2/3 of their time on administrative tasks and 1/3 with patients, he said.
AI can reduce time spent on marketing, supply chain, emails, research and documentation of patient visits, allowing doctors to focus more on patients than note-taking, Shah said.
“If we’re able to make our team members more productive, we’ll be better able to take care of patients and the population at large in our communities,” Shah said.
However, “every provider absolutely needs to check the notes and really look for things that are inaccurate and modify them or delete them,” said Dr. Kanan Modhwadia, a psychiatrist and medical director of behavioral health and psychiatric emergency room services at Northwestern Medicine Central DuPage Hospital.
:quality(70)/cloudfront-us-east-1.images.arcpublishing.com/shawmedia/Z6UJZIZEMFHHPISSW4NQAE2CME.jpeg)
Modhwadia said AI in psychiatry is in very early stages.
“Psychiatry is really driven by human connections and the relationship part, which AI can’t do. So that is a bit of a gap,” Modhwadia said. “But there are a lot of therapy platforms coming up.”
Silver Cross also recently started using AI in its electrophysiology lab to help improve treatments for patients with abnormal heart rhythms, including atrial fibrillation and ventricular tachycardia.
In 30 to 60 seconds, AI transforms ultrasound images into highly detailed 3D cardiac maps, leading to shorter procedure times, increased precision, and less exposure to radiation.
Dr. R. Kannan Mutharasan, cardiologist and medical director of the Bluhm Cardiovascular Institute at Northwestern Medicine Palos Hospital, and medical director of Health Network Development at Northwestern Medicine Bluhm Cardiovascular Institute, feels AI can transform patient care.
AI-powered stethoscopes can classify heart murmurs; AI-powered ECGs and echocardiograms can improve diagnostics, Mutharasan said.
Future AI may predict the risk of atrial fibrillation or a weak heart, Mutharasan said.
AI programs for clinicians, grounded in all the latest medical literature, can help providers give accurate, timely care because “it’s read every single thing humans ever read and seen every single case,” he said.
“As an experienced physician with well over two decades in cardiology care, I have not seen everything out there,” Mutharasan said. “AI has seen everything.”
St. Mary’s Hospital in Kankakee and Saint Joseph Medical Center in Joliet use an AI-powered device in the emergency department and the intensive care unit for rapid seizure detection – especially non-convulsive seizures – and continuous monitoring.
“Over 90% of seizures are subclinical, with no outward signs,” said Kristen Boswell, neuro-diagnostics supervisor at St. Mary’s Hospital.
:quality(70)/cloudfront-us-east-1.images.arcpublishing.com/shawmedia/PZ5HMM46QRC25F3KWOK2CW6EME.jpeg)
Boswell said the device allows staff to quickly make critical decisions on whether to treat the patient at the hospital or transfer to a hospital with neurointensive care.
And when dealing with the brain, “it’s just so critical to be able to do that,” Boswell said.
St. Mary’s in Kankakee was the first hospital in Illinois to use the device in children, too, Boswell said.
Dr. Rameez Alasadi, an interventional gastroenterologist with Silver Cross Medical Group, likened AI-enhanced colonoscopies to “a second set of eyes” that helps detect polyps.
Alasadi said the point of screening colonoscopies is to find and remove polyps before they turn into cancer.
“The higher the detection rate, the lower the risk of cancer,” Alasadi said.
Someday, AI might also highlight suspicious areas anywhere in the gastrointestinal tract for precise biopsying, he said.
“But we’re not there yet,” Alasadi said.
How AI should never be used in healthcare
Overrelying on AI or using it complacently is “where mistakes happen,” Modhwadia said.
“We should be thinking about what we’re doing and be very proactive,” Modhwadia said. “Because we’re ultimately dealing with people. It sounds very simple, but humans care for other people. AI responds. But it doesn’t reciprocate.”
Ironically, AI may restore the “sacred interaction” between patient and provider, Mutharasan said.
“Our vision for artificial intelligence is something that allows technology to completely fade away and elevate the human connection that is inherent in the care of the patient,” Mutharasan said. “I think that is an exciting possibility.”