- New tools like ChatGPT have sparked discussions about how AI could transform healthcare.
- While the potential is huge, healthcare systems are currently using AI for more mundane tasks.
- Leaders from four health systems shared how they use AI to manage emails or help doctors take notes.
The arrival of viral tools like ChatGPT has industry insiders imagining how artificial intelligence could revolutionize healthcare.
The chatbot, powered by a type of AI designed to produce text or images from written prompts, is said to be able to diagnose rare conditions and even pass the US medical licensing exam. This has raised questions – and fears – about the technology’s potential to replace doctors.
For now, however, US healthcare systems are using different forms of AI to solve more mundane and far less controversial problems.
They’re using AI to clean up and prioritize doctors’ inboxes, translate patient visit audio into summaries, and create staffing plans, so clinicians can spend less time at computers and more time with patients. These uses may not lend themselves to the plot of a science fiction novel, but hospital leaders predict they will have a huge impact on healthcare.
“This boring stuff right now is a big percentage of what a nurse and a doctor are dealing with,” said Scott Arnold, chief information officer at Tampa General Hospital.
“You’re not going to see robots providing care or making decisions for people. I think the big things where the AI ​​umbrella is going to really flourish will be helpful. It will help providers or nurses make decisions faster,” Arnold said. .
In turn, this could save a lot of money. A working paper released in January by experts at consulting firm McKinsey and Harvard University estimated that the adoption of AI by hospitals, physician groups and health insurers could reduce annual spending by healthcare in the United States from 200 to 360 billion dollars, or about 5% to 10. %.
Some health systems have also experimented with using AI to help diagnose disease. For example, some have used AI tools to double-check X-rays for signs of cancer that a busy radiologist might have overlooked. But health systems are generally cautious about deploying technology in clinical care, where the stakes are higher.
Health systems have a long way to go. Some say they are testing AI in smaller pilot programs to ensure results are accurate and data is protected.
Here’s how four health systems are using AI to tackle some of their biggest challenges.
Sutter Health uses AI to manage patient messages
Dr. Albert Chan, Director of Digital Health at Sutter Health.
Sutter Health
Sutter Health in California is exploring the use of AI to manage the volume of messages physicians receive from patients, said Dr. Albert Chan, the health system’s director of digital health.
The health system is working with Epic, an electronic health records company, on using AI to triage messages. This could potentially open up new possibilities, such as using AI to flag messages from patients who need urgent review by a clinician, Chan said.
He said doctors were inundated with emails, many of which could be handled by another member of staff.
“We know that if we can improve that, it’s a lot easier for doctors and patients to connect on things that really matter,” he said.
Sutter said it would be one of the first healthcare systems to use generative AI — the same technology that powers ChatGPT — to draft replies to patient messages, in partnership with Epic and Microsoft.
Chan said Sutter is also using AI in other ways, such as leveraging a tool from startup Ferrum Health to help radiologists find signs of lung cancer in X-rays.
AI helps UPMC document patient visits
Chris Carmody, Chief Technology Officer at UPMC.
UPMC
Chris Carmody, UPMC’s chief technology officer, said the Pittsburgh health system has been using AI for many years. UPMC notably helped develop the startup Abridge, which uses AI to record interactions between doctors and patients and then translate the audio into notes.
About 3,000 doctors at UPMC are testing Abridge and providing feedback to help improve the tool, Carmody said. He added that the goal is to integrate the tool into UPMC’s electronic health record so doctors don’t have to spend so much time documenting the patient’s visit.
Carmody said UPMC is also testing whether AI can scour unstructured data, such as discharge summaries, to identify patient symptoms so they can be used to inform how doctors care for patients. .
“Tools around AI, machine learning, generative AI are going to help us achieve this precision medicine that we’ve been talking about for years with technology,” he said.
Tampa General Hospital streamlines staff with AI
Scott Arnold, director of information at Tampa General Hospital.
Scott Arnold
Tampa General Hospital is working with Palantir Technologies to use AI to help nurse managers staff their units effectively, said Arnold, chief information officer.
He said Palantir’s tool analyzes available nurses and their skills and produces a staffing plan based on what nurse managers say the unit needs for each shift. He added that the tool has saved Tampa General hundreds of thousands of dollars.
The hospital is also working with Palantir to predict when hospital beds will be empty and ready to accept new patients so it can help patients get through the hospital faster. “It’s a big math problem that it allows us to solve quickly,” Arnold said.
Ardent Health Services frees up in-person nurses with an AI-powered tool
Marty Bonick, President and CEO of Ardent Health Services.
Ardent Health Services
Ardent Health Services, based in Nashville, Tennessee, has deployed an AI-powered tool from startup Care.ai to bring virtual nurses to a patient’s bedside to take care of the “tedious and time-consuming aspects of patient care “, said Marty Bonick, president and CEO of the health system.
The tool, which has been deployed at one of Ardent’s hospitals in New Mexico, allows virtual nurses to connect to rooms via clinical software and video to monitor patients, provide discharge instructions and document patient medications, Bonick said.
He said this frees up Ardent’s in-person caregivers to treat acute symptoms or administer medication.
A representative from Ardent said it plans to expand its work with Care.ai to implement virtual attendance to improve access to specialty care in rural communities.
Separately, Ardent has partnered with a Nashville venture capital studio called SwitchPoint Ventures to build AI companies to solve problems for healthcare providers. The first company they created was Winnow, which is designed to use predictive analytics to improve how healthcare systems recruit doctors.