ORLANDO, Florida—It's almost impossible to have a conversation at HIMSS24 without the topic of artificial intelligence or generative AI coming up.
Last year, at the HIMSS conference, an executive at electronic health record giant Epic noted that generative AI was at "peak hype."
The industry has now shifted from hype to strategic implementation as healthcare organizations partner with tech companies to explore use cases. Many health systems are testing out AI to automate mundane administrative tasks, and building the business case that it will help alleviate burnout.
Hackensack Meridian Health, New Jersey’s largest health system, is working with Google Cloud to deploy generative AI solutions to automate manual and repetitive tasks and analyze large patient data sets to identify patterns and trends to aid clinical decision-making.
"We've taken a strategic approach. We have various AI-enabled capabilities in production as pilots and under development," Robert Garrett, CEO of Hackensack Meridian Health, said during Tuesday's keynote speech at the HIMSS conference. "AI-driven chatbots are helping enhance the patient experience. We're using responsible AI with humans always in the loop."
The health system is developing AI solutions to help radiologists prioritize review of critical cases. "This can be a true game changer considering that a radiologist reviews up to 200 images every day," Garrett said.
Hackensack Meridian also is using AI to help physicians detect advanced kidney disease sooner to help delay dialysis treatment and even the need for transplants. "Another pilot is helping us optimize OR scheduling to fill in gaps that we know can lead to delayed care for patients. At the Hackensack Meridian School of Medicine, our students are learning how to integrate AI into their training and to prioritize ethics and patient safeguards."
He added that the health system takes a "sandbox approach" to develop AI platforms with companies like Google and other tech firms.
In the past two years, companies have been testing out "hundreds of thousands" of proof of concepts with generative AI and most did not make it into production, noted Matt Renner, president of North America and global startups at Google Cloud, who joined Garrett on stage for a discussion about AI innovation. "Many of these proof of concepts just didn't make the best business case. It wasn't going to make sense to invest to take that to production."
Healthcare organizations are trying to make the business case for AI to improve productivity among the workforce which helps to tackle broader issues around clinician shortages and burnout, Renner noted.
Working with Google Cloud, HCA Healthcare tapped generative AI tools to help streamline nurse hand-offs.
"They had a problem with the nurses and the hand-offs between nurses on a daily basis and a lot of burnout relative to that. In the HCA Healthcare system every day, there were 20,000 hand-offs between nurses from one shift to the next. One of the challenges is all the data they have from that day on the patients that they then pass to the next nurse. This is a very onerous process. When done with paper that will take anywhere from an hour to 90 minutes of their time," Renner said.
HCA created a virtual assistant to gather that data throughout the day and then turn it around into productive data for the next shift, which ultimately reduced time spent on that task by 80%, he noted. That enables nurses to spend more time with patients and helps to reduce burnout.
In the past year, health tech companies have rapidly rolled out gen AI tools with a specific focus on tackling administrative burdens for healthcare workers. EHR companies like Epic and athenahealth have developed new AI-powered tools to help summarize medical documentation and streamline paperwork. Oracle, which owns Cerner, also plans to launch an AI-based clinical assistant. At HIMSS this week, Symplr, an enterprise healthcare operations software company, announced it's working with Amazon Web Services to develop AI assistants to help automate complex workflows and mundane tasks for healthcare workers.
"The biggest cause of burnout has been the fact that clinicians can't spend enough time at the bedside," Garrett said. "They are on the electronic health record doing admissions, discharge notes and charting. That's a huge breakthrough for this technology to get those clinicians back to the patient's bedside. I do think this technology, in addition to enhancing productivity, will also reduce turnover."
Mayo Clinic has been using Google's AI solutions to improve radiation therapy planning for cancer care. Mayo worked with the tech giant to create and validate an algorithm that can automate healthy tissue and organs from tumors. The algorithm will boost patient outcomes and reduce the time it takes to plan radiation treatment, the health system said.
Conversations around AI have matured
As AI development rapidly accelerates, the focus at the C-suite level has shifted to foundational issues like data management and governance strategies.
"It comes down to data sets; the accuracy of the data, where the data is generated from in terms of governance," Garrett said. "I think governance and organization where you're overseeing AI and the development of AI algorithms has to be centralized. I like to think that the use cases, the pilots and the innovation is decentralized; it's at the care units, it's at the various care sites, but governance needs to be centralized."
He added, "At Hackensack Meridian, we take governance seriously. We have a centralized steering committee that applies principles and makes sure that as pilots develop and as use cases are developed that the information is accurate, that patients' privacy is protected and to make sure that the data set, that enterprise-wide data set, is appropriate. There's an operational steering committee that exists. I think we were also one of the first health systems to have a board committee dedicated to data governance."
The key is to support innovation but have standards in place, he said.
"We want to make sure that as the innovators are putting ideas forward, as they're solving for some of the problems, that we have consistency with data," Garrett noted.
The conversations about AI and generative AI have noticeably shifted in the past year, executives said at HIMSS this week.
"Last year when the gen AI storm hit, there was an incredible amount of excitement and people were trying to figure out what the strategy is. A lot of that has matured," said Kalyan Pamarthy, group product manager, Vertex AI Search for Google Cloud, where he focuses on building generative AI and search capabilities.
Organizations that invested in high-quality data platforms are now ready for this "big moment of AI," Pamarthy noted.
"They're seeing very quick results in the use cases. A lot of other customers have had to go through their journey, which is interesting because the analytics world has always known that you need good data to get good dashboards. The same also happens to be true for AI. We're seeing some of our early customers, who invested in healthcare data engineering platforms, are starting to see some very quick turnaround to get that set up," he said.
He added, "The quality of the AI comes down to the quality of the data."
Healthcare organizations also are keeping their eyes on potential regulation at the federal level. As AI technology rapidly advances, federal lawmakers will struggle to keep pace with regulations, Garrett noted during his keynote.
"Governance should emanate from the healthcare sector itself. And, we can work hand in glove with regulators in a collaborative partnership. The regulators are looking to us to put good safeguards in place," he said. "A good example of that is the executive order recently signed by President Biden and then the creation of a public-private group to determine how to implement safeguards."
The concept of "responsible AI" is dominating conversations as health systems, doctors, payers and other organizations weigh how to deploy AI solutions to ensure patient safety and privacy as well as output accuracy while also considering issues around bias or inequities.
AI hallucinations, when AI systems produce incorrect or misleading information, is a critical concern, especially in healthcare.
Renner said AI developers are working to reduce hallucinations by focusing on the concepts of factuality and grounding. Google, for example, ensures that its AI cites its sources or can connect its output to verifiable sources of information.
"If you think about it from a medical perspective, it's a classic second opinion. It could be a specific medical journal, it could be a CDC website. I think you'll continue to see this speed up and a lot more focus on factuality and grounding and other concepts that will come forward," Renner noted.
"We can't hold back this innovation. But, we must commit to deploying AI safely through effective governance. And, it can never replace human intervention and oversight. We must obtain and secure accurate data, protect patient privacy and commit to eliminating any potential for bias," Garrett said.
Predicting the future of AI innovation
Garrett said AI technology can help the healthcare industry "take another quantum leap forward,"
"Maybe AI will help us erase the 30-year gap in life expectancy between the richest and the poorest nations. And healthcare could become a universal right for all, instead of a privilege for some. And what if this stunning technology led to life-changing breakthroughs for the tens of millions suffering from mental illness and addiction? As we redefine and improve healthcare, we can never lose sight of our larger goal of enhancing the quality of people's lives in the most fundamental ways," he said during the HIMSS keynote.
There's no shortage of innovation coming from clinicians on the front lines of care who see ways to tap into AI to solve problems, Garrett noted.
"We have a tremendous waiting list. There's a backlog, and it's a good problem to have, but we can't keep up with demand out there because people want to solve their problems through these use cases," he said. "As long as you can create that culture that understands the potential and promotes innovation, while at the same time making sure that you have a pretty good centralized governance structure, I think you have a winning formula."
Renner cautioned that AI innovation in healthcare is a "marathon, not a sprint."
Healthcare organizations are finding more pertinent business cases to deploy AI and the AI models are getting more accurate, he said.
Garrett predicted that in a year, by HIMSS 2025, organizations will be accelerating AI implementations.
"I think some of these exciting use cases that are pilots now will start to roll out into practice. I think a year from now, there may be some regulations that the industry and regulators can agree upon," he noted. "As much progress has been made in AI over the last year or two, fasten your seat belts, everybody, this next year is going to be amazing."