More than 350,000 Americans suffer cardiac arrest events each year; however, only 2% of Americans are certified to perform CPR. Thus, in most cases when someone collapses, non-certified bystanders call 911, receive CPR instructions from dispatchers and do they best they can. Now, researchers have published a new study that shows an artificial intelligence-powered CPR coaching agent can outperform 911 dispatchers in guiding bystanders through CPR. “If AI is going to earn its place in medicine, it should start by helping people save the person right in front of them,” said study coauthor John Ayers, Head of AI at the UC San Diego Altman Clinical and Translational Research Institute. The findings highlight an opportunity to study AI’s role in delivering guideline-based instructions—as a complement, not a replacement, for human responders. “This is about supporting people in high-stakes situations where human judgment is essential,” said Christopher Horvat, M.D., Director of Medical Emergency Response Teams at UPMC Children’s Hospital of Pittsburgh. “The goal is to raise the floor of performance, not to replace trained professionals.” AI-assisted CPR The study, published in JAMA Internal Medicine, first benchmarked popular AI models—including ChatGPT, Claude, Grok, Gemini, Llama and Mixtral—on CPR coaching, comparing AI-generated CPR instructions with dispatcher-provided guidance from 12 real 911 calls. Across different situations, like drowning or collapsing while jogging, and different patients, from toddlers to seniors, AI models performed well on the basics of CPR. On average, they scored 90% on essential steps, such as where to press on the chest and how fast to do it. Scores ranged from 79% (Gemini) to 97% (Grok and Claude). However, when it came to giving the best possible instructions to improve survival, performance dropped. On these more advanced steps, such as letting the chest fully rise between compressions, models averaged 70%—Llama at the lowest (61%) and ChatGPT at the highest (75%). This poor performance inspired the researchers to develop ChatCPR, an open-source AI agent for CPR coaching. They used a separate set of real 911 calls where dispatchers had already provided CPR instruction to compare with ChatCPR’s instructions. On basic steps, ChatCPR met 100% guideline steps, while dispatchers met 85% of the steps. For more advanced steps, the gap was even larger—ChatCPR scored 99% while dispatchers scored 63%. Specifically, ChatCPR excelled in patient assessment, chest compression depth and rate instructions, and recoil guidance areas where multitasking dispatchers most often faltered. “This wasn’t about style. It was about strict adherence to CPR guidelines where precision matters most,” said study co-author Davey Smith, M.D., Director of the Altman Clinical and Translational Research Institute at UC San Diego. “ChatCPR addressed elements that dispatchers, under the stress and complexity of real calls, sometimes missed, presenting a valuable opportunity to translate AI into real-world healthcare practice.” Future framework The researchers say careful, real-world testing is still needed. They want to ensure the system is safe, works in chaotic settings and is easy for people to follow. Toward that goal, the researchers made ChatCPR open and free for anyone to use, study, refine, test and improve. Clear regulatory frameworks will also be essential as the tool moves from research to real-world use. Today, bystanders have strong legal protections against civil and criminal prosecution when intervening to perform CPR. How these protections extend to AI-enabled CPR is a challenge that needs to be addressed. “Ultimately, our work grounds AI hype in life-or-death reality,” concluded Ayers. “The real promise is closing the deadly gap between a person collapsing and lifesaving care beginning.”
More than 350,000 Americans suffer cardiac arrest events each year; however, only 2% of Americans are certified to perform CPR. Thus, in most cases when someone collapses, non-certified bystanders call 911, receive CPR instructions from dispatchers and do they best they can.
Now, researchers have published a new study that shows an artificial intelligence-powered CPR coaching agent can outperform 911 dispatchers in guiding bystanders through CPR.
“If AI is going to earn its place in medicine, it should start by helping people save the person right in front of them,” said study coauthor John Ayers, Head of AI at the UC San Diego Altman Clinical and Translational Research Institute.
The findings highlight an opportunity to study AI’s role in delivering guideline-based instructions—as a complement, not a replacement, for human responders.
“This is about supporting people in high-stakes situations where human judgment is essential,” said Christopher Horvat, M.D., Director of Medical Emergency Response Teams at UPMC Children’s Hospital of Pittsburgh. “The goal is to raise the floor of performance, not to replace trained professionals.”
AI-assisted CPR
The study, published in JAMA Internal Medicine, first benchmarked popular AI models—including ChatGPT, Claude, Grok, Gemini, Llama and Mixtral—on CPR coaching, comparing AI-generated CPR instructions with dispatcher-provided guidance from 12 real 911 calls.
Across different situations, like drowning or collapsing while jogging, and different patients, from toddlers to seniors, AI models performed well on the basics of CPR. On average, they scored 90% on essential steps, such as where to press on the chest and how fast to do it. Scores ranged from 79% (Gemini) to 97% (Grok and Claude).
However, when it came to giving the best possible instructions to improve survival, performance dropped. On these more advanced steps, such as letting the chest fully rise between compressions, models averaged 70%—Llama at the lowest (61%) and ChatGPT at the highest (75%).
This poor performance inspired the researchers to develop ChatCPR, an open-source AI agent for CPR coaching.
They used a separate set of real 911 calls where dispatchers had already provided CPR instruction to compare with ChatCPR’s instructions. On basic steps, ChatCPR met 100% guideline steps, while dispatchers met 85% of the steps. For more advanced steps, the gap was even larger—ChatCPR scored 99% while dispatchers scored 63%.
Specifically, ChatCPR excelled in patient assessment, chest compression depth and rate instructions, and recoil guidance areas where multitasking dispatchers most often faltered.
“This wasn’t about style. It was about strict adherence to CPR guidelines where precision matters most,” said study co-author Davey Smith, M.D., Director of the Altman Clinical and Translational Research Institute at UC San Diego. “ChatCPR addressed elements that dispatchers, under the stress and complexity of real calls, sometimes missed, presenting a valuable opportunity to translate AI into real-world healthcare practice.”
Future framework
The researchers say careful, real-world testing is still needed. They want to ensure the system is safe, works in chaotic settings and is easy for people to follow. Toward that goal, the researchers made ChatCPR open and free for anyone to use, study, refine, test and improve.
Clear regulatory frameworks will also be essential as the tool moves from research to real-world use. Today, bystanders have strong legal protections against civil and criminal prosecution when intervening to perform CPR. How these protections extend to AI-enabled CPR is a challenge that needs to be addressed.
“Ultimately, our work grounds AI hype in life-or-death reality,” concluded Ayers. “The real promise is closing the deadly gap between a person collapsing and lifesaving care beginning.”