How ambient AI can improve the patient-provider relationship
Ambient AI technology is helping Ochsner Health improve patient-provider relationships by allowing clinicians to spend consultation time looking at their patients rather than a computer screen.
Some people view technology as a barrier to human connection. However, ambient AI is having the opposite effect on patient-provider relationships, according to Jason Hill, MD, MMM, MS, innovation officer at Ochsner Health.
Researchers estimate that physicians spend over 40% of consultation time looking at the EHR screen, limiting their ability to connect with patients.
"We've got workers with a tremendous amount of talent spending time typing words into a computer rather than providing care, and you've got patient encounters where patients don't feel connected to their doctor because there's a screen in between them," Hill said.
What's more, growing documentation requirements for quality reporting, patient satisfaction and reimbursement are placing additional strain on providers.
Ambient AI scribes can alleviate the clinical documentation burden by allowing providers to spend the visit looking at their patients rather than at a computer screen.
The technology drafts clinical documentation in seconds by applying machine learning to conversations. Clinicians review the note and make any necessary edits before the system records the information.
Ochsner has been exploring ambient AI's potential for several years. However, until the advent of transformer-based large language models like OpenAI's GPT, it was more cost-effective to hire human scribes than it was to tune an AI.
I think what's really unique about this technology is that this actually has the ability to enhance a human connection, which you don't see very often with technology.
Jason Hill, MD, MMM, MSInnovation officer, Ochsner Health
"When language modeling and ambient AI was able to make that process within an affordable price range for a health system to scale out, we were very interested," Hill said.
Selecting a vendor
After conducting an initial market review, Ochsner piloted two ambient AI tools from different companies. Ultimately, the health system inked an enterprise-wide partnership with health IT vendor DeepScribe.
In the initial launch of the tool across 60 physicians, Ochsner saw an adoption rate of 75% and ongoing time savings. What's more, patient satisfaction scores increased by 6%.
"I think what's really unique about this technology is that this actually has the ability to enhance a human connection, which you don't see very often with technology," said Hill.
According to Hill, the app's customization features played a large role in Ochsner's purchasing decision.
"We can set templates for specialties or service lines, but individual doctors have control and agency over different note outputs depending on the clinical encounter that they're taking place in," said Hill. "That was key for buy-in. If you can't meet doctors where they are for documentation needs, it doesn't work."
He also emphasized the role of strong vendor support in the purchasing decision. In addition to working with Ochsner support staff to design training modules, they've been shoulder-to-shoulder with clinicians to address concerns.
"Whenever clinicians have had a problem, they have flown people down here to help them out, or they've done work through Zoom or customization through the app," Hill said.
For instance, he noted that the vendor developed a new software module for Ochsner within about a month that allows providers to see EHR reminders within the ambient AI app.
"DeepScribe created a bridge to move that information from the EHR into the app so that when they're looking at the app about to start the visit, they can see that right there," Hill said.
First, providers receive training before they can use the ambient AI functionality. The education underscores that clinicians must review all ambient AI outputs. Once the AI creates the draft, the note is in a "pended" state, which requires action from the provider to be signed into the record.
"We also measure the amount of text generated by the ambient AI versus added by the provider, which gives us a surrogate marker on editing," Hill said.
The health system is scaling the pilot to 60 more clinicians, with plans to extend ambient AI to 600 Ochsner doctors by the end of the year. Ultimately, the health system's 4,700 physicians will have access to the tool.
While Hill expects that most Ochsner clinicians will adopt ambient AI for documentation, he pointed out that it might not be for all providers.
"There are some clinicians who don't have a lot of documentation requirements, or their documentation requirements are in some other system, but for the vast majority of our clinicians, this will probably be the way that we do documentation at Ochsner within a year," he said.
Hannah Nelson has been covering news related to health information technology and health data interoperability since 2020.
