Hospitals are testing AI in the Emergency Department, where clinicians face high rates of burnout and need tools to help them create a better medical record.
Generative AI is proving it's value in the doctor's office, where the doctor-patient encounter is usually structured and quiet. Now healthcare leaders want to apply that technology to the Emergency Department, where very little is controlled.
At Atlanta's Emory Healthcare, ED clinicians are using Abridge Inside for Emergency Medicine, a generative AI tool designed to organize, maintain and update the patient's medical record through the disjointed and often interrupted journey from admission to discharge.
Tricia Smith, MBBS, MPH, FACEP, an emergency physician at Emory University Hospital Midtown, says the technology has the potential to reduce time spent gathering disparate patient data, giving clinicians a more efficient path to diagnosis and treatment.
And that's a critical pain point in a hospital's most stressful environment.
"If you have a complete story on the first encounter that you have with the patient, you're able to streamline your workflow," Smith says. "It seems like a minute here, a minute there, but those things really add up in a chaotic environment."
Emory is one of a handful of health systems that have integrated the AI tool into the Epic EHR platform for ED use over the past three months. Abridge officials say the technology taps into Epic's ASAP module to capture the salient points of the conversation, identifying key words and different speakers, and create a medical record that a clinician can drop into and out of as needed.
One of the bigger challenges of the ED is the disparity of the information needed to treat a patient, especially one who may not be conscious or able to communicate. Clinicians in this environment are gathering data from paramedics and EMTs, family members, visual exams and whatever they can find that's already in the EMR. They're developing a medical record out of these unstructured pieces of data, while also stepping in and out of the room to deal with other concerns.
Smith says that workflow extends to the end of this journey as well, when the patient is either discharged or admitted. That medical record needs to be complete for the next care provider, either in the hospital or at home, as well as the patient's insurer and the hospital's revenue cycle department. And AI can help organize that process.
"I have to write these words in this order in this context so the patient can understand exactly what I mean, so the insurer can understand exactly what I mean," Smith points out. And right now we haven't hit that sweet spot for emergency care to make sure that the note has all of those check buttons matched just yet."
Smith says it's likely the technology will need to be fine-tuned as it spends more time in the ED, encountering as many different distractions and roadblocks as possible. That includes making sure the technology is really helping clinicians—burnout rates for this group are among the highest in the hospital setting, running from 50% to 70% in several recent studies.
Smith says ED clinicians face unique challenges, exacerbated by the environment. There are many different points of pressure on their cognitive workload, and it can easily be more difficult to gather and organize the right information than to make a diagnosis. The right tool for them isn't necessarily the one that will help make the diagnosis, but will help them on the path to that diagnosis.
After hearing all the stories about how AI is helping the clinician capture the patient conversation, Smith says she's grateful the C-Suite is looking beyond the doctor's office and into the ED.
"To be quite frank, we felt a little bit left out of the AI conversation," she says. "The majority of these generative AI tools that were being developed were developed for clinic settings, for office settings, where you have a 15-minute encounter. There's a back and forth and there's a neat and tidy conclusion all wrapped up within 15 minutes. But in the ED you're trying to make order out of chaos."
Eric Wicklund is the associate content manager and senior editor for Innovation at HealthLeaders.
KEY TAKEAWAYS
Burnout rates for clinicians in the ED range from 50% to 70%, according to several recent studies.
While AI is proving successful in helping doctors capture the patient conversation in an office setting, the technology needs to be refined to help clinicians in a chaotic setting like the ED, where they’re dealing with several challenges and trying to gather data from many different sources.
A handful of health systems are testing out an AI tool that can capture, assemble and update information on the fly, enabling clinicians to create a more complete medical record and hopefully improve outcomes down the line.