![<?echo $_SERVER['SERVER_NAME'];?>](/template/twentyseventeen/skin/images/header.jpg)
According to foreign media reports, researchers at the University of Washington have recently developed a new non-contact tool that can monitor people's heartbeat during sleep. A new technology for smart speakers or smartphones allows the device to detect wheezing without area breathing and initiate a call for help. The proof-of-concept tool was developed using a real, non-geographical breathing example captured from a 911 phone, which detected an area-free breathing event an average of 97% of the time outside of 20 feet (or 6 meters).
University of Washington Paul G. Shyam Gollakota, an associate professor and co-author of the Allen School of Computer Science & Engineering, said that many people have smart speakers in their homes. These devices have amazing features that we can use. We envision a non-contact system that continuously and passively monitors non-contact breathing events in the bedroom and alerts anyone nearby to CPR. If no response is made, the device will automatically dial 911.
According to data from emergency telephones, about 50% of patients with cardiac arrest have non-directional breathing, while patients with non-directional breathing tend to have greater chances of survival. Dr. Jacob Sunshine, assistant professor of anesthesiology and pain medicine at the University of Washington School of Medicine, co-corresponding author of the study, said: "This breathing occurs when the patient's oxygen level is very low, which is a wheezing sound on the throat. Its uniqueness makes it a good audio biomarker that can be used to identify a person with a cardiac arrest."
The researchers collected the voice of asexual breathing from a real phone call and collected 162 calls between 2009 and 2017, extracting 2.5 seconds of audio at the beginning of each asexual breath, for a total of 236 clips. The research team is on different smart devices: Amazon's Alexa, iPhone 5s and Samsung's Galaxy S4 capture these recordings and use a variety of machine learning techniques to boost the data set to 7,316 positive segments.
Justin Chan, a Ph.D. student at Allen School, said: "We play these examples at different distances to simulate what it would sound like if the patient was in a different position in the bedroom. We also added different disturbing sounds. Such as the sound of cats and dogs, the horn of cars, the sound of air conditioners, and the sounds you can usually hear at home."
For negative data sets, the team used audio data collected during an 83-hour sleep study to get 7305 sound samples. These clips contain typical sounds that people make while sleeping, such as snoring or obstructive sleep apnea. From these datasets, the team used machine learning to create a tool that detected non-directional breathing in 97% of cases when the smart device was placed 6 meters away from the speaker that produced the sound.
Next, the team tested the algorithm to make sure it didn't accidentally divide different types of breathing, such as snoring, into zoneless breathing. Chan said: "We don't want to send unnecessary alerts to emergency services or relatives, so it's important to reduce the false positive rate."
For sleep laboratory data, the algorithm incorrectly classifies the breathing sound as a breathing sound of 0.14% of the time. For individual audio clips, the false positive rate was about 0.22%, with volunteers recording themselves while sleeping in their own home. But when the team used the tool to classify something as a drone, the false positive rate of the two tests fell to 0% only when it detected two different events at least 10 seconds apart.
The team envisions that the algorithm can be run like an app or run like a Alexa on a smart speaker or smartphone while people sleep. Gollakota said: "This can be run locally on the processor included in Alexa. It runs in real time, so you don't need to store anything or send anything to the cloud." The researchers plan to split the sound through UW Life sciences commercialize this technology.
Carbon Steel Products,Carbon Steel Parts,Carbon Steel Precision Casting,Cnc Precision Casting Parts
Dongying Jingcheng Metal Products Co.,Ltd , https://www.dyjcmetal.com