As part of a goal to return 10 million clinical hours to clinicians by 2028, Philadelphia-based Jefferson Health is piloting an ambient AI documentation tool for nursing units — a largely unprecedented goal for health system AI projects.
The system operates 33 hospital campuses and approximately 700 care sites throughout Pennsylvania and New Jersey. In September, the organization announced the time-saving plan, which also broadly aims to improve patient care and outcomes, cut administrative burden and enhance operational efficiency.
Jefferson Health President Baligh Yehia, MD, told Becker’s the system wants to save each clinician an average of one hour of administrative work per day, which adds up to 10 million hours that can be used for patient-facing care. The saved time is measured through signal data, which are the AI tool logs, as well as in-person observation to verify the data.
An early aspect of this project featured ambient listening technology available to Jefferson’s physicians, advanced practice providers and physical therapists working in specific settings, including ambulatory sites.
So far, the system has logged about 350,000 hours, or 3.5%, thanks to the ambient technology, according to Colleen Mallozzi, RN, senior vice president and chief nursing informatics officer at Jefferson Health.
Ms. Mallozzi and Dr. Yehia told Becker’s they expect other AI-focused technologies, including a documentation tool that could be made available to its 14,000 nurses and other staff, will expedite the system’s progress toward 10 million hours.
The ambient AI pilot for nurses
Based on the success of physicians using ambient listening technology for documentation, Jefferson is taking a similar — but not identical — approach for nurses.
Physicians document their work through treatment decisions and care plans in EHRs; while nurses document in standardized flow sheets with drop-down menus and checklists. That fundamental difference influenced how Jefferson is testing ambient AI for nursing.
At three nursing units at Abington (Pa.) Hospital, a 667-bed regional referral and teaching hospital, care teams are piloting ambient AI technology. The tool translates spoken language into structured flow sheet documentation.
After gaining a patient’s consent, the nurses, patient care technicians, patient care coordinators and nurse managers on those three units verbalize their assessments. The ambient technology ingests the care conversations and delivers a structured output for nursing flow sheets.
The first hurdle for Jefferson? Nurses do not always verbally communicate what they are doing.
“Imagine if you had to go throughout your day just verbalizing everything you were doing. How awkward would that be?” Ms. Mallozzi said. “Nurses are used to quietly going about their assessments, doing three different tasks at once, and now they have to verbalize it. So this was a big change. There’s an awkwardness to get over, to figure out how to have that conversational care with your patients and verbalize some of the things you were doing.”
‘We have camps’: The adoption challenge
About 30% of nurses are hesitant to use the technology, Ms. Mallozzi said.
Jefferson hosted a panel discussion about the technology April 22 for about 400 nurse leaders to query staff members who are taking part in the pilot. Common concerns raised by staff included legal ramifications, environmental effects, potential over-reliance and feeling like they are “performing” for the AI rather than focusing on the patient.
She said newer-to-practice nurses are more hesitant to verbalize their assessments, likely because they are “still getting their feet grounded in their practice.”
“It’s been a little polarizing. We have camps,” Ms. Mallozzi said, adding that Jefferson plans to make the model opt-in after the pilot, similar to the rollout for the physicians’ ambient documentation tool.
Results so far
The pilot is in its early stages, so nursing-specific outcomes are still emerging. However, Ms. Mallozzi said the ambient tool is changing the way nurses work.
“The technology is amazing, but this is about transforming the way that nurses practice, the way that they engage with their patients: to verbalize their care,” Ms. Mallozzi said. “It’s incredible, because it’s helpful for the patients to understand what you’re going through, what you’re thinking through, that critical thinking aspect.”
She raised an unexpected benefit: the added educational value for nursing students at the teaching hospital. Nursing students do not have to guess at the clinical reasoning behind their supervisors’ decisions — the verbalization requirement makes that thinking explicit and audible in real time.
For documentation, Jefferson Health leaders are predicting similar results from its ambient tool for physicians and APPs. That tool has decreased time spent on EHRs by about 17%, as well as 16% less time spent on charts outside of work — sometimes called “pajama time” — and 40% less time to close a chart, Dr. Yehia said.
On a larger scale, he said Jefferson is saving clinicians an average of 15 to 20 minutes per day.
Jefferson’s broader AI strategy and governance structure
The system has identified and implemented more than 100 AI-powered solutions, including ones focused on care delivery, revenue cycle operations and back-office tasks.
Each idea and technology is funneled through Jefferson Health’s AI Center for Excellence, which launched in 2022. A key component of the system’s AI governance structure is usability.
“The biggest underline for me is this [AI technology] is being implemented by our clinical staff, for them and by them,” Ms. Mallozzi said. “We aren’t just making broad stroke decisions for what they’re going to use. … We run [technology] by them, we let them poke holes in it and then we pilot it with them.”
If an AI tool does not click with clinicians, it does not make it to a systemwide launch, she said.
Jefferson’s near-future AI plans include agentic technologies, such as voice agents for follow-up phone calls and appointment scheduling.
We are not “implementing technology for the sake of implementing technology,” Ms. Mallozzi said. “It truly is grounded in returning time to care: care for yourself, care for your patients. I think it’s really important that we look at, are you getting time to take a breath after a complex patient experience or a death of a patient? Are you getting time to take your lunch? The time may be given back for that bedside care and time with patients, but if our clinicians are caring for themselves, they’re going to be better caregivers for other individuals as well.”
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.