Researchers from the University of Washington have developed a system that uses smart speakers and machine learning algorithms to monitor heartbeats without the need for physical contact. The system was described in a study published in Communications Biology.

When a person sits within two feet of a smart speaker, the system sends out inaudible sounds from the speaker that reflect off a person’s chest and back to the speaker. Using a “beam-forming algorithm,” the system can identify and monitor individual heartbeats, whether regular or irregular. Smart speakers marketed by Amazon (Alexa), Google (Nest) and other tech companies use multiple microphones that can isolate heartbeats and detect changes in the returned sound.

In a University of Washington news release, a School of Medicine professor pointed out that cardiac arrythmias are more common than some other more recognized heart conditions and can cause major morbidities, such as strokes. The new system has the potential to vastly improve detection of these abnormalities using devices that are already in people’s homes.