A new artificial intelligence system has been installed in Magen David Adom’s 101 Emergency Dispatch Center to help EMTs and paramedics get the information they need, so help can arrive faster.The system, developed with the voice networking and media processing solutions company AudioCodes, works in Hebrew.
According to MDA, it is the first of its kind in the world specifically intended for use in an emergency dispatch center. The system transcribes the call on the 101 line in real-time and is helpful to the dispatch center in cases where there is a disturbance in the quality of the call.
“In an emergency, if someone is dying or there is a lot of shouting or other voices around the caller, it is harder for the call taker to understand what has happened immediately,” explained Dr. Eli Jaffe, MDA’s deputy director-general for community outreach. The new system helps solve this challenge by identifying keywords for medical treatment and emphasizing them to bring them to the attention of the emergency medical technician (EMT), who can then give the most relevant advice to the caller and the dispatch teams. These could be words like shooting, car accident, chest pains, unconscious, etc.
“Try to imagine there is shooting around us,” Jaffe described. “We can hear the shooting and the panic and the anxiety. I call, and the call taker asks me questions. I want to give him the information immediately, but I need to talk very fast and very quietly because I don’t want the shooter to find me. And the call taker must try to understand what is happening.”
Jaffe offered as an example the 2008 Mercaz Harav terror attack, in which a yeshiva student called MDA while hiding under a table. It was hard to understand him, the outreach official recalled. But the new system would have immediately typed the keywords, making it more efficient and faster for a dispatch team to understand the situation and arrive at the scene.
How does the AI system work?
MDA and AudioCodes have been working on the system for two years. It now transcribes the entire conversation for Israel’s national emergency medical service in real-time, so questions don’t have to be asked a second time. In addition, it has learned the key phrases connected to medicine and emergencies to provide an EMT with the overall picture and enable rapid, professional decision-making.
“The accuracy of the AI transcribing solutions has significantly improved and can sometimes prove to be better than human ability – for example, the ability to respond to and analyze speech that is whispered or of a low sound quality – and presenting them to the call takers in MDA accurately and quickly,” said Shabtai Adlersberg, president and CEO of AudioCodes.
Currently, the system is operating in Hebrew, but Jaffe said it would be able to work in all “Google” languages soon.
“This is just the beginning,” Jaffe said, noting that MDA is looking at additional AI solutions to improve its service. “We want to save more time and save more lives.”