If 2023 was the year of generative AI hype in healthcare, 2024 also will be the year of generative AI hype in healthcare.
“We’ll be in the hype cycle for a while given its newness,” said Sara Vaezy, chief strategy and digital officer at Renton, Washington-based health system Providence. “The reality is the availability of enterprise-grade large language models is small because a lot of the startup companies are building tiny features because they need to bring in revenues immediately or don’t have the data to scale their models.”
Most health systems are looking at generative AI from a use-case perspective in areas where obvious solutions exist for obvious problems such as documentation and messaging. Rohit Chandra, chief digital officer at Cleveland Clinic. said it’s too early for widespread adoption beyond those smaller use cases.
“There are more unknowns than knowns,” Chandra said. “But there are a few problems that are discreet and specific enough that I think that it will drive organizations to adopt solutions.”
Health executives are excited about those use cases and the impact they will have. An early December survey of leaders at hospitals, home health providers and ambulatory clinics found 58% of healthcare organizations expect to adopt a generative AI solution in 2024. Research company KLAS, which performed the survey, found only 25% of organizations adopted one in 2023.
Here's how four health systems plan to adopt generative AI in 2024.
Providence is working with electronic health record vendor Epic Systems as well as internally to reduce the number of inbox messages clinicians receive from patients, Vaezy said.
“It’s becoming a bigger and bigger burden from a time management standpoint for our clinicians,” Vaezy said.
Vaezy said the physician communication capability at Providence aims to triage messages. In some cases, physicians don’t need to answer the patient’s query and it can be done through a chatbot. It’s also developing a capability that if a message does need to be sent to a physician, it is put into the clinician’s workflow in a way that doesn’t waste time.
The health system also is creating an “internal GPT” model to use for non-clinical use cases, she said. Nurse managers could use it to direct bed capacity in various units throughout the hospital.
An area where Vaezy doesn’t expect a lot of progress in 2024 is in clinical use cases. “Clinical is the area where we're being the most careful and the most measured,” she said.
Baptist Health is another early adopter of Epic’s generative AI communication tool to reduce the number of patient messages to clinicians. Chief Digital and Information Officer Aaron Miri said the system's physicians are working with technologists to build prompts that clinicians can use when responding to patients. Epic CEO Judy Faulkner said at an early December event that the company was getting good feedback on the capability.
Similar to Providence, Jacksonville, Florida-based Baptist is building private large language models. In one use case, the health system asked employees how they felt about their work during an all-team meeting and used AI to summarize the sentiments, Miri said.
“We’re using AI not just for operational efficiencies but for staff development,” Miri said.
Baptist also set up an AI institute to ensure the technology is being legally, ethically and thoughtfully deployed, which will help as the health system receives vendor AI pitches, Miri said. “A lot of what these vendors are selling is smoke and mirrors,” he said.
Houston-based Memorial Hermann plans to use AI for nurse engagement and retention, said chief strategy officer Feby Abraham. The company recently implemented an AI solution from startup Laudio that allows managers to reduce repetitive tasks and increase efficiencies for nurses. Abraham envisions using generative AI to make all clinician lives easier.
“Today, a physician or a nurse spends so much of their time on work that may be deemed as less of a value add,” Abraham said. “It’s not seeing patients or engaging in clinical care. How can they use AI to augment what they’re doing. We’re partnering with a lot of companies to do exactly that.”
Abraham said he sees particular use for generative AI around improving patient-physician interactions and answering questions around billing. But clinical workflows can be augmented too, he said.
“Maybe it’s an automated interaction or even looking for the type of care that should be recommended to a patient based on their specific data,” Abraham said.
Cleveland Clinic is also looking to improve patient and provider experiences through generative AI, such as using it to help translate medical terminology for patients, scheduling appointments and refilling medications.
For providers, Chandra said generative AI can be used to streamline documentation. “Physicians are spending 10 hours per week doing a variety of administrative tasks,” Chandra said. “If we can reduce that from 10 to seven to five to two, that’s great.”
Cleveland Clinic also is focused on ensuring the right guardrails are in place. In early December, the health system co-founded the AI Alliance, a community of researchers, developers and organizational leaders aiming to develop and achieve safe and responsible AI.