Abridge expanded its generative AI tools for medical documentation into the inpatient setting, the company announced Wednesday.
The new offering extends Abridge's current functionality for outpatient and emergency medicine to support inpatient care, integrating into existing Epic workflows. Abridge Inside for Inpatient was co-developed with electronic health record giant Epic as part of the company's participation in Epic's Workshop program. The program features third-party vendors that are co-developing, with Epic, technologies that are piloted and refined before wider release.
Abridge Inside for Inpatient is tailored specifically to the unique Epic inpatient workflow, Abridge said, and features seamless integration with Haiku to enables users to leverage SmartPhrases and templates for review and rapid note closure. The new inpatient offering also gives clinicians the option to choose from history and physical, progress notes and consult note types, according to the company.
Epic will automatically map the generated note into the selected note type. In an upcoming update, the progress note will incorporate prior note data from Epic, Abridge said.
Abridge Inside for Inpatient also facilitates pre-charting, post-charting and switching between patients.
Pittsburgh-based Abridge, one of Fierce Healthcare's Fierce 15 of 2024 honorees, uses AI to increase the speed and accuracy of medical note-taking, leveraging a proprietary data set derived from more than 1.5 million medical encounters. The company's AI converts a patient-clinician conversation into a structured clinical note draft in real time and integrates it seamlessly into the EHR system.
Abridge's technology has now been deployed in more than 100 health systems.
In January, the company launched a new gen AI product for emergency care, and it's currently in use at several health systems including Deaconess Health System, Emory Healthcare, Johns Hopkins Medicine and UChicago Medicine. The emergency medicine gen AI tool was designed in collaboration with ED clinicians.
Inpatient workflows are very different from outpatient visits. Multiple clinicians have multiple touch points with a patient every hour.
"The inpatient space is equally, if not more, complicated than the emergency room," said Shiv Rao, M.D., a cardiologist who launched Abridge in 2018 and now serves as CEO. "Any given patient is seeing any number of different doctors on the primary team, and then there's consultants who are rounding as well. There could be trainees who are rounding on those patients as well. It's just a very hectic sort of situation in the inpatient world."
"Being able to work well inside those workflows requires not only a deep Epic integration, but it requires a lot of work on our side to create notes that can really meet the mark over there," Rao said in an interview with Fierce Healthcare about the new capabilities.
Sutter Health, as an early health system partner, is piloting the inpatient gen AI product.
"Abridge Inside automatically transforms bedside conversations into structured Epic notes, empowering inpatient clinicians who deliver the intricately balanced coordinated care for admitted patients to devote more time to patient care,” Veena Jones, M.D., vice president and chief medical information officer at Sutter Health, said in a statement.
"Seamless Epic integration into a variety of note types makes it easy to use, and our admitting hospitalists, in particular, are thrilled," Jones said.
Richard Liu, M.D., a hospitalist at Sutter Mills-Peninsula Medical Center in Burlingame, California, said the inpatient gen AI tool "calms some of the chaos of delivering inpatient care."
"Just by holding my phone and pressing a button, I know the information shared by each patient is captured and available for documentation in their chart," Liu said.
Expanding its gen AI tools to the inpatient setting opens up an entirely new surface area for innovation, Rao noted.
"Those workflows, on the innovation side of things, are very complicated, and there's a real need to have as much context as possible and create the best possible notes," he said, adding, "Revenue cycle is a really important aspect of what we do. We are, in essence, a mid-revenue cycle company, because the notes that we generate essentially function as bills for insurance companies."
Rao said the new capabilities and features represent how the company "is going millions of miles deep" on developing gen AI tools for different care settings and different clinical workflows.
"Our mission in many ways, from a product perspective, is to go millions of miles deep on all the different ways that we can help unburden clinicians from clerical work," he said. "One way we translate that internally is that we try to go millions of miles deep on all the different settings where care is delivered, all the different specialties. That might also mean going deep on all the different spoken languages as we have multilingual technology."
The company also unveiled a new gen-AI-powered feature to streamline the outpatient orders workflow through an early pilot in Epic's Workshop program.
The outpatient orders feature allows medications mentioned during patient encounters to be surfaced directly inside Epic, enabling clinicians to rapidly place orders based on the conversation—saving time, reducing cognitive burden and minimizing duplicate work, according to the company.
“Once I finish the visit, the medications I talked through are already surfaced in Epic—and that’s a huge help in a busy clinic,” said Mary Kirby, M.D., a family medicine physician at Coastal Carolina.
"We're starting to extend into workflows that are just beyond the documentation aspect of what a doctor does," Rao said. "If I'm seeing you in a clinic, I open up Haiku, which is Epic's mobile app, I hit record on Abridge. We would have a conversation when I hit stop, my note should be there inside Epic, inside the desktop medical record. But now, not only is the note there, but my orders can be templated. My orders were automatically extracted ambiently from the conversation that I had with you, when I said you should get a CT scan. All those orders are already templated for me to similarly trust and verify."
Abridge plans to continue adding features to support imaging, labs and additional order types to help clinicians close the loop on more types of follow-up care discussed during the patient conversation.
The AI scribe market is booming, with investors pouring millions into the sector. Abridge picked up $250 million in series D funding back in February at a reported valuation of $2.75 billion. The company has raised more than $460 million, according to public records.
Clinical documentation has been a "wedge" for AI to make inroads in healthcare, Rao noted.
"We've been able to scale faster than most companies in any industry, or any company in any industry. It's been a very exciting few years," he said.
As more health systems and provider adopted AI scribe and ambient documentation technologies, companies will need to expand their capabilities and create more value.
"The co-development that we've been doing with Epic in the Workshop program for order entry is a really good example of what comes next. What also comes from the conversation? Well, orders are discussed. So can we actually get into order entry? There will be an opportunity to go really deep to make sure that we're performing well across all the different medications and procedures and referrals out there, and make sure that we're highly accurate and deeply integrated," he said.
As AI in healthcare evolves, many companies are focused on applying the technology to back-office or front-office tasks or "jobs to be done," as Rao said.
"As long as we're responsible about thinking about which are the low-stakes tasks, we'll probably find any number of opportunities that are relatively low hanging in terms of the ability to create real value very quickly," he said. "I think over the coming years though, we'll start to move from those lower-stakes workflows into potentially higher-stakes workflows that AI can impact, for example, clinical decision support, and as we move into those new use cases, we'll obviously have to be very responsible and thoughtful as we deploy and scale this technology."
For now, though, Rao noted there are plenty of documentation and workflow tasks for Abridge to tackle. A study published in 2022 found that primary care clinicians need 26.7 hours per day to follow national recommended guidelines for preventive care, chronic disease care and acute care while seeing an average number of patients.
"At Abridge, we have our work cut out for us. There's any number of different clerical tasks and workflows that we can impact with AI. We work with over 100 health systems across the country. What we've done here with inpatient and orders, we can really start to land and expand from there into new workflows or new care settings," he said.