Alexa will be able to determine cardiac arrest, listening to a person's breath

If you are lucky and you have to stop the heart, the chances are slim. Of the 400 000 people who are faced with this every year in one country, survive less than 6%. This is the main cause of natural death. However, the probability of survival will be much greater if render immediate assistance: immediate resuscitation can double or triple the odds. The new tool uses intelligent microphone speakers or smart phone to detect signals timely can improve survival rates.

How to save a person in cardiac arrest?

The system, developed by scientists from the University of Washington, uses machine learning to determine the so-called gasping, breathing (also known as agonal breathing), which manifests itself in a person who does not have enough air. This is an early warning sign that appears in half of all cases of heart failure.

The scientists trained the system using the recording agonistic respiration obtained by calls to emergency care in King County, Washington. They used the 729 calls, which gave a total of 82 hours of recordings. They then trained the system to other sounds, which can also be heard in the room, such as snoring or sleep muttering in order to weed out any false positives. Use two different sets of records: sleep sounds gathered 35 volunteers, and the sounds of the 12 patients who participated in a sleep study as suffering from snoring and sleep apnea. Recent Entries produce sounds similar to agonal breathing, helping to improve the accuracy of the instrument.

"When we tested it on your system, we found that the level of false positives in the group of volunteers was 0, 2%, while in a sleep study - 0, 1%," says Justin Chang, who led the study.

The system could correctly identify agonal breathing in 97% of cases at a distance of 6 meters.

While the tool is being proof of concept, so before its release in the general public for many years, although the researchers plan to eventually commercialize it. Much remains to be tested. Most likely, the system will automatically call an ambulance, but warn the person about it so that he could cancel the call.

What do you think, how many lives will save such a system? Tell us in our chat in the telegram.