Across the globe, up to 7 million people die of a heart attack yearly. A new application of artificial intelligence could save lives. Works with smart loudspeakers.
Almost 500,000 Americans die each year from a heart attack. For German medical societies in the same period, 65,000 cases are reported. Every third person affected is younger than 65 and often does not know the risk factors before the event. Since many heart attacks occur outside the hospital, medical assistants are missing for revival.
Researchers from the University of Washington have therefore developed a new program that allows people to follow without contact, such as sleeping. You're writing, many of the heartbreak occur in the bedroom. Thanks to modern technology, the help could be in the eye: intelligent speakers such as Google Home and Amazon's Alex, as well as newer smart phones, can analyze respiratory sounds and, in the worst case, even automatically get help.
Digital assistants should give life save
Background: In 50% of all patients with heart failure, so-called snap-breathing occurs. "This type of breathing occurs when a patient has a very low oxygen content," explains Jacob Sunshine. Anesthesiologist and pain medicine at the University of Washington Medical School adds: "It's a gloomy mood and the noise is so characteristic that audio biomarkers can be used to determine if there is a heart failure."
But how should it go home? "Many people have smart loudspeakers and these devices have incredible features that we can use," adds Shyam Gollakota. He works at the School of Computer Science and Engineering Paul G. Allen at the University of Washington. "We anticipate a contactless system in which the bedroom is continually and passively monitored for the critical sound of breathing, and all people nearby are warned of system messages." If they do not respond, it's possible to report to the Emergency Center.
Calling from the emergency call center
It took several steps to complete. The researchers initially collected breath sounds in emergency calls from Emergency Services in Seattle. Since heart failure patients are largely unaware, they liked to record their breath sounds by holding the phone in the patient's mouth. Between 2009 and 2017, the team collected exactly 162 phone calls that actually included patients with cardiac arrest.
Audio files are played back and then recorded with Amazon Alexa, iPhone 5s and Samsung Galaxy S4. "These examples were conducted at different intervals to simulate how it sounds when the patient is in different places in the bedroom," says Justin Chan of Washington University. "We've also added a lot of noise, from cats and dogs, to car trumpets, to air conditioning, to things you usually hear at home."
Reduce error rate – with artificial intelligence
In the assessment, researchers used artificial intelligence. First, breathing sounds in patients known to have cardiac arrest are stored in the database. The noise was added. Using algorithms, scientists used programs to search for specific samples in the signal. This software was then tested with the right data, with results that are constantly being optimized, the keyword "Machine Learning". The result was a tool that correctly recognized breakdown in 97% of all cases. The distance between the device and the patient in the simulation was a maximum of 6 meters.
The researchers then tested their algorithm to make sure that other sleep sounds are not misunderstood as snap-breathing. Finally, false positive signals would quickly lead to the end of all events. Depending on the setting, the rate of false positive results was between 0.14% and 0.22%. As soon as your software evaluates sounds as a snap-breathing event, if there were two identical readings within ten seconds, the rate dropped to 0%.
From laboratory to application
After making a feasibility study, it should move quickly to the application. The team imagines that the algorithm can work as an application or Alex's "skill". The application should be activated before bedtime. This circumstance is a possible obstacle: After all, digital assistants are criticized because they have too many ears. Since many risky patients are not known as such, it is difficult to benefit from the new technology. So it is known that patients are at risk of heart failure.
More about medicine: