High tech innovations at the EMS Copenhagen dispatch center are giving people experiencing out-of-hospital cardiac arrest (OHCA) a better run for a longer life.
And there’s data to prove it.
The technology, streaming video and artificial intelligence, is part of international initiatives—such as NG911 in North America and the NG112 standard in Europe—to aid emergency dispatchers in further assisting callers, bystanders, field responders, and incident commanders.
The tools featured in the EMS World conference presentation (Sept. 19) “High Tech Innovations in Dispatch” focused on their application to CPR, although next steps in emergency dispatch innovation could include stroke and sepsis identification and instructions.
“Time is of the essence [in cardiac arrest], and dispatch has the biggest impact on survival,” said EMS Copenhagen CEO Freddy Lippert, also a founding member of the Global Resuscitation Alliance, who was among four presenters during the hour-long virtual session.
In Copenhagen, Denmark, emergency calls coming into Region Hovedstadens Akutberedskab (roughly “EMS in the Capital Region of Denmark”) are screened with the location verified and, if medical, the call is forwarded to an emergency medical dispatcher to determine the appropriate response.1
When the 112 call (the equivalent of the U.S. 911 emergency number) sounds like a cardiac arrest, the medical dispatchers send the caller a text message that contains a hyperlink. When the calltaker clicks the link, the medical dispatcher can view the scene through the caller’s cellphone camera, live and in real time. The medical dispatchers guide the bystanders to perform CPR until the arrival of the ambulance and use the nearest automated external defibrillator (AED).2
There are 20,789 AEDs available at just as many locations in Denmark and an active Civilian First Responder System.
Preliminary data of 700 calls involving cardiac arrest from a video pilot program showed that response was changed in 30% of the calls incorporating the live streaming. The percentage was not evenly split, with 10% requiring a higher level of response and 20% requiring a lower level once the emergency dispatcher was able to view the situation.
Most dispatchers (98%) in the pilot favored live streaming, Lippert said, indicating that it contributed valuable information about the OHCA patient, the physical setting, and the bystanders’ response.
An earlier study (2014) conducted prior to broader use of video streaming by EMS Copenhagen examined OHCA recognition from audio files of calls to its emergency medical dispatch center (EMDC). For the study, a machine learning framework was trained to recognize cardiac arrest from the recorded calls. The machine learning framework was compared to the actual recognition and time-to-recognition of cardiac arrest by medical dispatchers.
Results showed that time-to-recognition was significantly shorter for the machine learning framework compared to the dispatchers (median 44 seconds vs. 54 seconds, p < 0.001).3
However, although machine learning can potentially improve OHCA survival, researchers concluded that the technology “should not be used as a stand-alone tool that can independently dispatch ambulances but could act as a supplement to dispatchers’ decision-making processes based on standard operating procedures, algorithms and personal experiences.”4
Moving beyond machine learning is the pursuit of artificial intelligence applications in emergency dispatch, as discussed in the same session by Catherine Counts, Research and Quality Improvement Manager with Seattle Medic One and University of Washington School of Medicine.
“It’s [artificial intelligence] an area we can expect the most improvement in identifying every cardiac arrest as soon as possible,” said Counts, an investigator in multiple OHCA and EMS studies.
AI is able identify the potential of cardiac arrest faster through indicators described by the caller, such as “lips are blue” or “he’s making gasping sounds.” Earlier recognition leads to more rapid and accurate dispatch, faster hands on chest, and the correct application of high-performance CPR.
AI does not preclude the necessity of a human, Counts said.
“They work in tandem,” she said. “AI can determine how much faster an emergency dispatcher can trigger a CA alert.”
Although used interchangeably, AI and machine learning are not the same.
Machine learning examines and compares data to find common patterns. AI solves problems. It has been defined as a computer system able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.5
Copenhagen EMS is split into five regions countrywide. The EMDC receives about 130,000 calls a year to its 112 emergency number. Denmark is member of an international alliance to promote international standards-based Next Generation emergency communications frameworks in North America and Europe. Under both frameworks, requests for emergency help are handled in an Internet Protocol-based multimedia environment, as opposed to the voice- and landline-centric frameworks of the past.6
1 Gates H. “EMS Around the World: Denmark Prides Itself on Innovation.” EMS World. 2019; Oct. 29. https://www.emsworld.com/article/1223380/ems-around-world-denmark-prides-itself-innovation (accessed Sept. 21, 2020).
2 See note 1.
3 Blomberg S, Folke F, Ersboll A, Sayre M, Counts C, Lippert F. “Machine learning as a supportive took to recognize cardiac arrest in emergency calls.” Resuscitation. 2019; May 1. https://www.resuscitationjournal.com/article/S0300-9572(18)30975-4/fulltext (accessed Sept. 21, 2020).
4 See note 3.
5 Nicholson C. “Artificial Intelligence vs. Machine Learning vs. Deep Learning.” Pathfinder. https://wiki.pathmind.com/ai-vs-machine-learning-vs-deep-learning (accessed Sept. 22, 2020).
6 Fontes B. “NENA-Initiated National Alliance Will Work to Accelerate NG Emergency Communications Globally. 2019; Nov. 11. https://www.nena.org/news/477525/NENA-Initiated-International-Alliance-Will-Work-to-Accelerate-NG-Emergency-Communications-Globally.htm (accessed Sept. 22, 2020).