AI Comes for EMS

Artificial intelligence (AI) couldn’t be any hotter, any bigger a force in our lives, in our world, and across our healthcare system. It’s remaking industries, transforming jobs and professions, offering us new electronic friends and therapists, and using up massive amounts of energy. AI is also generating controversy and fears, including among frontline EMS personnel, but that’s not going to slow it down.

Donnie Woodyard, Jr., MAML, NRP, WP-C, veteran paramedic, EMS educator and executive director of the Interstate Commission for EMS Personnel Practice in Washington, DC, has written and presented extensively about AI. He emphasizes that the topic is not new but dates back to the early days of computer science.

“What has changed is the ability for it to reach down to the lowest levels and for the common person to be able to access it,” he told EMS World. “And now it’s in this exponential expansion, where it can be used in so many aspects of our lives.”

Woodyard believes that the field of EMS is at an inflection point relative to AI as it struggles to clarify its own identity.

“Is EMS healthcare? Or is EMS transport to healthcare?” he asked. “If EMS wants to identify itself truly as healthcare, healthcare now demands integration with technology and with AI and, more importantly, with the consumers of healthcare who are demanding that it integrate AI and technology into a comprehensive user experience.”

Our healthcare system increasingly uses AI to help with patient communication, the recording of patient encounters, voice recognition, billing, real-time data capture and more, he explained. “This is the point where I believe that if EMS does not rapidly figure this out, it will remain in the category of primarily transport to healthcare.”

The infusion of AI into the EMS ambulance so far has largely involved the electronic patient care record (ePCR). But for Woodyard, the models that have been implemented, by and large, are pretty basic, relative to the current state of the art in AI technology.

“Yes, there is value in having a tool that helps with reporting,” he said. “But I hear some in the industry asking, ‘Is this it?’ And my answer is, ‘No. What you’re being given right now are the crumbs of AI, the leftover table scraps.’”

This is not what AI has the potential to do, Woodyard explained. “It is not how the rest of healthcare, business, technology, or national defense is beginning to leverage AI, and frankly, it is not even the full potential of AI-assisted documentation.”

Saving Time for Crews

Although the degree to which EMS agencies and their staff have considered or embraced the possibilities of AI varies widely between sites, its boosters emphasize potential opportunities to give more time back to the ambulance crew by automating routine tasks and improving efficiency. They also point to its potential for enhancing dispatch triage, with more dynamic deployment of ambulances, even redirecting them based on real-time traffic data.

Could AI systems learn to make faster stroke diagnoses based on speech patterns, or the identification of other emergent clinical trends? Could image recognition make it possible to enter data into the chart just from a photo taken of a medication bottle or blood bag? What other kinds of clinical decision support, reminders and suggestions, predictive alerts and the like—many of which are finding their way into hospital practice—might be looming for EMS?

Rick Hines, Fire Chief, Ketchikan Fire Department, Ketchikan, Alaska, views AI as another tool for EMS teams, a tool still in its infancy but with huge upsides. It’s having a rollout similar to other big changes in EMS practice over the years, such as the introduction of ultrasound or ventilators or defibrillators, now stocked in ambulances.

There is a learning curve, he said, “But hopefully it’s going to make our jobs more efficient and make us better able to provide better care to Mr. and Mrs. Smith, the people who need our services.”

Hines, who oversees fire and EMS operations for the department, including community paramedicine initiatives and system-level improvements focused on patient care, response efficiency, and workforce sustainability, extolls AI’s ability to capture data for the department’s new EMS reporting platform.

“We’ve also had the ability to use other AI tools to [compile] data. I think that it’s important when asking for money, or asking for new equipment, that we should be able to back up that request with data,” he said.

Hines’ EMS crews like that they can dictate their ePCR narratives out loud, including vital signs and the like.

“The AI platform will take all of that information and put it into the correct box or the correct area in their narrative,” he said. They still need to go back and check and double check, but that takes a lot less time than having to enter it from scratch.

“The other thing is that any EMS agency struggles with billing,” he said. “Your reimbursement is based on what you put in your narrative in a category that the insurance company understands. So, you can train it to get optimal reimbursement.”

Overcoming Fears

Ed Bauter, a paramedic, CEO of Overrun EMS, and a PhD candidate in health sciences, started doing research on AI for his advanced coursework. He started working on a master’s degree in 2022, just as the company OpenAI got a big infusion of capital funding.

As Bauter had more conversations with people in the field about AI and its applications, he came to recognize how people really don’t know what it means.

“We’ve been using Spellcheck and Grammarly for AI assistance for years without any problem,” he told EMS World. “But the reason why I think definitions are important is that you have people saying they’re afraid of it. They’re worried about large language models becoming autonomous.”

At this point, the frontline provider’s interaction with AI may largely be unseen, built into the process. But he thinks field staff will soon see a greater efficiency and reductions in workload—which could translate to less fatigue and burnout.

“For administrators, you can ask it questions like: How can I improve my EM system? How do the best EMS services improve response time, or develop high yield with their QA/QI programs?” he explained.

Zachery Alvey is a full-time paramedic for Ute Pass Regional Ambulance District, Woodland Park, Colorado, which serves a mountainous rural area of about 600 square miles, near Pike’s Peak, with three ambulances. He acknowledges that he is not a “tech guy.” But talking with Woodyard helped spark his interest in augmented reality and AI and large language models.

“I started thinking maybe I need to jump on board before this all passes me by,” he said. “It looks like this is the direction we’re moving, so I started embracing the use of this technology. This is something that we in EMS don’t even know how to embrace yet. I think there’s a lot of fear around the various formats of AI.

“For editing the charts, just recording our conversations and implementing it into an ePCR system, there are a couple of products out there that are really great—physician-developed, 27-bit encrypted, meeting all HIPAA standards.”

These systems can filter out the background conversation and other noise and highlight the main focus of the interaction, compiling it together and using it to create a well-organized narrative with a clear pattern of diagnoses and recommendations.

“If we could, for example, do some X-rays, or even a CT, while in the field, with an algorithm programmed to look for certain hot points, we could reach a preliminary diagnosis,” Alvey said.
“Is this a large vessel occlusion? Is this a bleed? Maybe we would skip stopping at the nearest rural hospital and instead get that patient to definitive medical care quicker.”

Or another example, high-acuity, low-frequency pediatric cases, which are challenging to paramedics.

“So we think about using AI in the back of my ambulance where it’s watching, it’s listening, and it can actually start piecing together the probabilities of what’s going on with this kid. I see potential benefits of diagnoses made in the field with AI, since we just can’t process the information as fast as that electronic machine can,” he said.

Alvey has also found it valuable in his work as an EMS educator. “AI in the EMS classroom has been fantastic because it helps me translate concepts that are very difficult. I can use it to augment those situations and help my students gain a better understanding.” And that has been reflected in their test scores.

He sees AI offering particular benefits in rural areas such as where he works. “The implementation of AI now bridges gaps for the provider shortages we have. I see it expanding and saving lives,” he said. “I say it’s not AI that will replace us in our jobs. It’s the kid who knows how to use it.”

The Future is Now

AI is going to change the physical ambulance and what it’s stocked with, as well as how EMS crews deploy, respond and transport, Woodyard said. Take, for example, Waymo and other autonomous vehicles or autonomous flight.

“Right now, I encounter EMS systems putting on blinders and ignoring the possibilities of autonomous vehicles,” he said. “But I tell them, you’d better not do that, because this will intersect your world very quickly.” Autonomous vehicles are safer and will improve access, he explained.

Woodyard’s message to frontline users is this: “Don’t resist the technology. Educate yourself and understand how it’s going to be [used]. When we’re talking about enterprise-level AI, which is able to write with accuracy, they should start playing with it. I think every medical professional will need operational fluency in AI. Use these models enough to understand their inherent strengths and weaknesses.”

EMS is the only part of the healthcare system that takes care of patients across a whole spectrum of environments, Woodyard said. “And the environment holds unique clues to the true origins of the patient’s problems. EMS finds the patient at the house, in the woods, by the side of the road. And clues about that environment could help unlock the mysteries of some of what’s going on with this patient. AI should be able to capture many of those clues, both the subtle clues and the obvious ones, and bring that into the continuum of care.”

It's a wonderful tool, he said. “But what’s coming next is going to be mind-blowing. And that is agentic AI,” which refers to autonomous AI systems that are capable of pursuing complex, multi-step goals with only minimal human supervision. “That is really going to the next level, taking all kinds of inputs, of information, and making high-level recommendations for care. That is going to become front and center in the next 12 to 24 months.”

For Hines, the future will also include some kind of video recording, a body cam, and a camera mounted in the back of the ambulance. It will be able to extract from the recordings the information to generate the patient narrative for the chart, once it’s trained to do so.

“To even recognize a procedure being done and whether it was successful or not. And that information goes to the hospital, to the insurer—additional key information in a different medium than typed words,” he said. “We’re also thinking of items like Google’s smart [VR] glasses that you could train to look at a patient and get a pretty accurate height and weight measurement. I can look at a suspected fracture and know if it’s really a fracture. Or I can look at an EKG and the Google glasses will be able to read it and transmit the data automatically to the hospital.”

All of which is aimed at shortening door to treatment times and solidifying what’s going on in the care. “It’s evolving at such a fast rate,” Hines said.