Ochsner Health going all-in on ambient AI, taps DeepScribe to deploy tech for 4,700 physicians

Ochsner Health is going all-in on ambient AI voice technology for its thousands of doctors and has inked an enterprise agreement to roll out DeepScribe's technology across its health system.

The health system piloted DeepScribe's technolology, along with another AI scribe vendor, earlier this year and has seen notable results for patient satisfaction and provider adoption, executives said.

Ochsner Health, a large nonprofit healthcare provider in the Gulf South, will eventually roll out the technology to its 4,700 employed and affiliated physicians across its 46 hospitals and 370 health and urgent care centers. 

DeepScribe was founded in 2017 by Akilesh Bapu, Matthew Ko and Kairui Zeng to use AI to revamp how clinicians document care and tackle the challenges of medical documentation. Its ambient AI clinical platform automatically produces complete, accurate documentation and delivers real-time insights to physicians in the moment of care, according to the company. DeepScribe claims it built its latest technology, HealAI, on the largest database of patient conversations. HealAI has been trained to manage the most complex cases, with customized workflows for each medical specialty, according to executives.

The health tech company works with 1,200 provider organizations and has raised about $35.2 million to date from investors.

DeepScribe's customization "studio" and flexibility was a key driver of physician adoption during the pilot, according to Jason Hill, M.D., Innovation Officer at Ochsner Health.

"We did a side-by-side pilot with DeepScribe and one other solution and decided ultimately to settle on DeepScribe, and it was for a number of reasons. We felt DeepScribe was just a great partner. They are willing to dive in with us with the problems that face our organization to provide quick solutions to them," Hill said in an interview.

"We found that their customization system is really key to overcoming that first hurdle, which is adoption because if you can't overcome the cost hurdle and the adoption hurdle, nothing else matters. If you can't find value in the solution, and if you can't get providers to use the solution, nothing else that you do downstream of that is worthwhile," said Hill, who also is board-certified in internal medicine and clinical informatics.

The company's customization capabilities and deep Epic electronic health system (EHR) integration made it easier for Ochsner to deploy ambient AI with specialty-specific workflows that accommodates physician preferences while also aligning with Ochsner-specific requirements, Hill noted. "The result is consistently high-quality notes and high clinician adoption," he said.

DeepScribe captures clinical conversations between doctors and patients during medical visits using advanced speech recognition technology and then automatically generates clinical notes in the medical record. Clinicians can choose from more than 50 options to personalize their notes using DeepScribe's technology, according to executives.

Hill added, "What makes DeepScribe unique is that it's highly flexible and it has a customization studio that they do through their app, and it allows doctors to customize their workflows. If you've talked to any doctor, every doctor is special and has their own special note template, so being able to customize that is very key to being able to scale adoption."

The DeepScribe pilot project started with around 50 Ochsner Health providers in primary care as well as eight other medical sub-specialties including palliative care, oncology and nephrology, Hill said.

"This was a really representative sample of the type of environment that we're going into," Matthew Ko, DeepScribe COO and co-founder, said in an interview. "The pilot was originally scoped for 20 providers, but because we saw such early success, we wanted to test it out in more clinical settings, so we extended it to around 50." 

Ochsner Health leaders and DeepScribe evaluated the initial pilot based on several metrics, including the amount of time providers saved on medical documentation, adoption of the technology adoption and patient satisfaction.

Ochsner Health reported a 78% clinician adoption rate during the initial launch and documentation time was reduced by around 75%, according to the organizations. "That's turning the time providers are spending on notes from hours to minutes," Ko said.

“Before using DeepScribe, I spent at least two to three hours a day preparing for visits and then going back and editing notes based on patient conversations,” Terrance Wickman, M.D., an Ochsner Health nephrologist, said. “Now, I have been able to streamline that into three to four minutes per note. So documentation takes me about a quarter of the time.”

While there is a lot of hype around ambient AI in healthcare, Hill says the technology's ability to unburden clinicians from tedious data entry marks a game-changing shift.

"I can tell you, I've implemented EHRs at around 40 hospitals. I've never actually had a doctor send me a message, like a video of themselves telling me how amazing something was in 40 EHR implementations. When I implemented this, I had doctors sending me videos of themselves telling me how happy they are with this product, and how this has changed their lives," he said. "That was an 'aha' moment to me." 

Use of the medical AI scribe technology also increased patient satisfaction, according to results of the pilot.

"We were able to improve the patient satisfaction from 89% to around 96% during the pilot comparing the users that didn't use DeepScribe to the users that did," Ko noted.

When evaluating Ochsner Health clinicians using DeepScribe, 96% patients stated they are likely to recommend the provider. 

"When I looked at the comments from the patients, it was very clear. Patients said things like, 'My doctor is now more engaged with me.' 'I feel like my doctor is listening to me.' All of those comments that we're hearing from patients are a marker that their doctors are more engaged, which tells me that this is a value. Patient loyalty and patient satisfaction and engagement all leads to downstream benefits for the health system," Hill said.

Ko said DeepScribe developed a purpose-built large language model for medical conversations. 

"One of the things that we're finding in the market is that it's really easy to see a demo of an ambient solution. What's hard is building a product that is battle tested and something that works in the wild, when it's encountering all the unexpected scenarios. And those scenarios change specialty to specialty," Ko said. "What we've been able to do is really lean into the ability to actually customize the notes and customize the model to accommodate for not only the specialty-specific preference, and requirements, but also the individual provider-specific preferences." 

The partnership with DeepScribe sets the stage for further expansion of its technology throughout Ochsner Health.

"We're planning to scale this out across our system over the next year. Our first scaled target is our primary care service line, which is around 300 providers, physicians and APPs (advanced practice providers). That's going to start in August. Very closely after that, we're going to be scaling to our orthopedic service line, which is around 75 providers, physicians and APPs," Hill said. "That will probably get us to the end of the year. We're looking beyond 2025 to evaluate other service lines and so small pilots in other service lines to see where those value equations lie."

"You're never going to get 100% adoption," Hill said. "You don't structure innovation to the laggards. You structure innovation to the majority of your people, the 80% of people that you do feel will adopt. I think that's where we're really trying to find where those value streams exist because this is not a free technology."

He continued, "We have to figure out a way that we can drive value for this. And maybe that's value in the ability to synchronize and standardized documentation across the service line. Maybe that's a value to help drive our procedure lists to be able to do more procedures and less time sitting and typing notes. Maybe that's value that's driving our primary care doctors to feel better at the end of their day and to help recruit more and retain more."

He added, "Those are each separate value streams that you'd have to think through when you adopt this. I think each of those is going to be somewhat service line-dependent. But, if we create a flexible enough system, we can then help to drive that service line value wherever we need to."

Beyond just a technology deployment, DeepScribe and Ochsner Health are co-developing ambient AI solutions that will best serve the health system's clinical and organizational needs, executives said.

Through its collaboration with Ochsner, DeepScribe is focused on building a business case for ambient AI clinical solutions by improving the quality of documentation or building new functionalities, Ko noted.

"The pace at which the market is evolving is incredible validation for the need for these types of solutions. I think it's still in the very early innings. It's very easy to kind of see and feel this visceral experience when you look at the magic that a lot of these LLMs can do. But in order to actually turn them into battle-tested products that work every single time, it requires a whole different perspective," Ko said. 

"I think a lot of that is going to come out in the next few months where we're really going to be able to see if these tools can evolve from just being a quality of life tool that helps saves physician time to really being a tool that can actually produce financial outcomes for the organization," he said.

As a health system that embraces tech innovation, Ochsner is evaluating these AI medical scribe technologies for benefits beyond eliminating provider burnout and time savings. "They're looking at deploying these technologies as far as how do you actually build the ambient intelligence layer that sits across these types of solutions and building functionality there. This is a true partnership, it's not just a customer relationship. We've been really working at the elbow with Ochsner leadership to figure out what other use cases they want to build along that spectrum," Ko said.

As Ochsner Health continues to deploy the technology, Hill and his team will evaluate provider adoption as well as outcome metrics like RVU (relative value units) production.

"We might be looking at documentation excellence, or the ability to make sure that our documentation meets the standards of compliance that we set out. So instead of now having to train 1,000 doctors on documentation standards and compliance, we can train a tool and we can fit that tool for compliance rather than trying to teach everybody," he said. "It might be looking at retention and physician joy. And, lastly patient satisfaction. Patient satisfaction is often a surrogate of doctor satisfaction."