- Generative AI and its role in healthcare took center stage at the HIMSS Global Health Conference in Chicago this week.
- There is particular excitement about using AI to reduce paperwork and burdensome office tasks so that healthcare professionals can spend more time with patients.
- Companies like Microsoft, Google, and Amazon are partnering with healthcare organizations to implement healthcare solutions using generative AI.
HIMSS conference attendees walk the exhibition floor
Debates about artificial intelligence and its role in healthcare took center stage at the HIMSS Global Health Conference in Chicago this week, where more than 35,000 physicians, other healthcare professionals, executives and engineers gathered to discuss the latest advances in field of health and technology.
Companies like Microsoft, Google, and Amazon prominently advertised new AI healthcare applications on booths across a sprawling exhibit floor, and panels of experts answered questions about how the technology can be used to address industry-wide challenges like understaffing and physician burnout.
Many healthcare organizations and companies have been using AI in various capacities for years, but a subset known as generative AI exploded into public consciousness late last year when Microsoft-backed OpenAI launched its new viral chatbot called ChatGPT. can use quite complicated requests from end users to generate text or images.
Just as generative AI has captured the attention of the general public, it has also captivated the medical community.
Artificial intelligence was the focus of the HIMSS conference keynote address, and HIMSS CEO Hal Wolf prefaced the discussion by revealing that he asked ChatGPT how to solve global healthcare challenges. The Healthcare Information and Management Systems Society, or HIMSS, holds the conference annually.
Wolf posed the question to ChatGPT as a joke, but David Rhew, global chief medical officer at Microsoft, told CNBC in an interview that generative AI could indeed be “transformative” for solving big problems in healthcare.
“The opportunity to apply these great language models and AI in clinical workflows is tremendous, and we need to do it responsibly,” he said.
For Rhew, that means starting with “high-impact, low-risk” uses for the technology, such as simplifying administrative tasks.
Developing diagnostic or patient-facing generative AI applications carries higher risk as it poses significant regulatory issues for companies, academics, and federal agencies like the Food and Drug Administration to work on. Rhew said he thinks of AI as if the healthcare industry had just been introduced in a car, while none of the stop signs, traffic lights or roads have been created yet.
“We still have to figure out how to do this together,” she said.
HIMSS CEO Hal Wolf speaks at the HIMSS conference
But meanwhile, administrative or ‘back office’ tasks require less regulatory oversight and there is a real need for efficient solutions, as clerical work is often taxing for physicians.
A study funded by the American Medical Association in 2016 found that for every hour a doctor spent with a patient, he spent two more on administrative work. The study said doctors also tend to spend an extra hour or two on clerical work outside of office hours.
Similarly, in 2017, the Journal of the Association of American Medical Colleges published a survey in which respondents said that approximately 24% of their working hours are spent on administrative tasks. More than two-thirds of physicians surveyed reported that administrative responsibilities “adversely affect their ability to provide high-quality care.”
HIMSS attendees told CNBC they believe Generative AI could help with these tasks.
On Monday, Microsoft announced an expanded partnership with Epic Systems, a healthcare software company that helps hospitals and other healthcare systems store, share and access electronic health records. More than 160 million people use Epic’s MyChart software, which gives patients direct access to health information and the care team.
Epic’s first application of AI technology automatically generates draft responses to messages doctors receive from patients via MyChart. Physicians do not have to use the suggested draft at all, but they save time if they choose to edit or send it.
Seth Hain, senior vice president of research and development at Epic, told CNBC in an interview that AI could serve as an impactful hypothesis generation tool for clinicians in the future. He said they’ll be able to ask patient-specific questions like: What do you think I should look at next regarding this issue?
Peter Lee, corporate vice president of research and incubations at Microsoft, told CNBC that a first look at Epic’s AI developments brought tears to his eyes.
“It blew me away,” she said.
Microsoft’s speech recognition subsidiary Nuance Communications also announced a clinical notes application called DAX Express ahead of HIMSS in March. DAX Express aims to help reduce physician administrative burdens by automatically drafting a clinical note seconds after a patient visit.
In a live demo at HIMSS, Nuance previewed future plans and showcased DAX Express features, which were met with gasps and cheers from a few doctors, nurses and healthcare professionals in the room.
More than 35,000 people attended the HIMSS conference in 2023
Other companies are also working to use generative AI to reduce administrative burdens.
Amazon Web Services on Monday announced an expanded partnership with Philips, a health technology company based in the Netherlands. AWS has already supported many of Philips’ existing AI and cloud-based initiatives, such as those that help radiologists analyze medical scans and images faster even from home.
But Monday’s announcement means that Philips will also use generative AI technology from AWS to streamline its clinical workflows and further enhance its imaging capabilities.
“What’s most exciting is the fact that we’re approaching a precipice where we have this tipping point, where we make the right thing easy,” Shez Partovi, Chief Innovation and Strategy Officer at Philips, told CNBC. an interview. “And right now, in most technology, the right thing is just a few clicks away.”
Partovi said all the little tasks doctors have to complete are like “death from 1,000 cuts,” so using AI to solve administrative challenges can have a real impact on doctors’ quality of life.
3M Health Information Systems also announced on Tuesday that it is also working with machine learning and generative AI from Amazon Web Services to help reduce physicians’ administrative workload. 3M HIS supports a conversational AI platform used by more than 300,000 physicians, and the company said in a release that AWS technology will make it easier for physicians to automate and complete accurate medical notes in the electronic health record.
Similarly, last week Google Cloud announced a Claims Acceleration Suite that uses artificial intelligence to streamline claims processing and health insurance pre-authorization.
According to the Centers for Medicare & Medicaid Services, the current pre-authorization process takes an average of 10 days. Google’s AI will help alleviate some of that administrative burden for providers by converting unstructured data that appears in images, PDFs, or other medical records into a more easily digestible structured format.
“They actually require a human being to go in there and take that data and feed it back into the system for review,” Amy Waldron, director of global health plans strategy and solutions at Google Cloud, said during a briefing with reporters. at HIMSS. “Which, to me, makes absolutely no sense given that someone has to take the time to enter all that rich data, and we have AI that can unlock that value.”
Generative AI has “tremendous” potential to improve administrative efficiency in healthcare, said Microsoft’s Rhew. But as healthcare and technology companies continue to make more sophisticated advances, industry leaders, regulators and community academics will need to ensure that generative AI is fair and does no harm to communities.
The technology is vulnerable to bias and discrimination if it is trained on health data that does not adequately represent a patient population, which could ultimately lead to poor treatment decisions or plans.
Consequently, Rhew said, there is a collective responsibility to figure out how to implement AI with care.
“It’s a transformative technology,” he said, “but we need to figure out how to do it responsibly.”