Amazon Alexa could pick up on a patient in cardiac arrest

Smart speakers could one day pick up if someone is suffering a cardiac arrest and call emergency services, research suggests.

A cardiac arrest occurs when the heart suddenly stops pumping blood around the body, which often causes a sufferer to produce a ‘guttural, gasping noise’.

Researchers have developed a new tool for devices such as an Amazon Alexa that detects this disrupted breathing and contacts emergency services.

When tested on real calls, the tool picked up on cardiac arrests 97 per cent of the time, even when the gasping sound was 20ft (6m) away from the receiver. 

The University of Washington researchers hope the ‘compelling’ technology will one day be added to smart speakers as standard to help ‘catch more patients in time’. 

Smart speakers could one day pick up if someone is suffering a cardiac arrest

Study author Dr Shyam Gollakota said: ‘A lot of people have smart speakers in their homes and these devices have amazing capabilities we can take advantage of. 

‘We envision a contactless system that works by continuously and passively monitoring the bedroom for an agonal breathing event.’

He added that it alerts anyone nearby to come provide CPR. However, the scientists did not explain how it does this. 

‘And then if there’s no response, the device can automatically call 911,’ Dr Gollakota added. The tool can filter out background sounds.

The research was led by Justin Chan, a PhD student in the department of computer science and engineering.  

Almost 500,000 Americans die each year from a cardiac arrest, the researchers wrote in the journal npj Digital Medicine.

And the condition kills 100,000 Britons annually, according to Arrhythmia Alliance. 

Study author Dr Jacob Sunshine, assistant professor of anesthesiology and pain medicine, said: ‘Cardiac arrests are a very common way for people to die and right now many of them can go unwitnessed. 

‘Part of what makes this technology so compelling is that it could help us catch more patients in time for them to be treated.’ 

Immediate CPR can double or even triple a patient’s chance of survival, however, it relies on a bystander being willing to help. 

UK research suggests nearly a third of adults (30 per cent) would not perform CPR if someone was suffering a cardiac arrest. 

This is said to be largely due to bystanders lacking in confidence they could perform CPR correctly.

In the US, nearly 383,000 cardiac arrests occur outside of hospital every year, of which 88 per take place at home, American Heart Association statistics show. 

The bedroom is one of the most likely places for a cardiac arrest to occur, with many sufferers being alone or with their partners sleeping besides them, the researchers claim.

Smart speakers could therefore be well placed to pick up on the gasping noises a patient makes, which is known as agonal breathing.  

Agonal breathing occurs in around half of people suffering cardiac arrests, according to emergency-services data. 

‘This kind of breathing happens when a patient experiences really low oxygen levels,’ Dr Sunshine said.

‘It’s sort of a guttural gasping noise and its uniqueness makes it a good audio biomarker to use to identify if someone is experiencing a cardiac arrest.’

To put the concept to the test, the researchers collected audio clips of agonal breathing made during 162 calls to Seattle’s Emergency Medical Services between 2009 and 2017.  

Bystanders recorded the sounds by putting phones up to the mouths of the patients, who were often unconscious.   


A cardiac arrest occurs when the heart suddenly stops pumping blood around the body, which is usually due to a problem with electrical signals in the organ.

This causes the brain to be starved of oxygen, which results in sufferers not breathing and losing consciousness.

In the UK, more than 30,000 cardiac arrests occur a year outside of hospital, compared to over 356,000 in the US.

Cardiac arrests are different to heart attacks, with the latter occurring when blood supply to the heart muscle is cut off due to a clot in one of the coronary arteries. 

Common causes include heart attacks, heart disease and heart muscle inflammation.

Drug overdose and losing a large amount of blood can also be to blame.

Giving an electric shock through the chest wall via a defibrillator can start the heart again. 

In the meantime, CPR can keep oxygen circulating around the body.

The researchers captured the recordings on different smart devices, ranging from an Amazon Alexa to an iPhone 5s and a Samsung Galaxy S4.

They then used various machine-learning techniques to boost the dataset to 7,316 audio clips.

‘We played these examples at different distances to simulate what it would sound like if it the patient was at different places in the bedroom,’ Mr Chan said. 

‘We also added different interfering sounds such as cats and dogs, cars honking, air conditioning, things that you might normally hear in a home.’

For the control data, the researchers collected 83 hours of audio from sleep studies, which were made up of noises like snoring and obstructive sleep apnoea.

Results revealed the tool detected agonal breathing 97 per cent of the time when the device was placed up to 6m away from a speaker that was generating the sound.

The team then tested the technology to ensure it would not accidentally class different types of breathing, like snoring, as agonal.

‘We don’t want to alert either emergency services or loved ones unnecessarily, so it’s important we reduce our false positive rate,’ Mr Chan said.

The tool incorrectly categorised a sound as agonal breathing just 0.14 per cent of the time. 

The false positive rate was around 0.22 per cent when tested on audio clips that volunteers recorded while sleeping in their own homes.  

The researchers hope the smart-speaker tool will one day be available as an app or will run ‘automatically’ on devices while we sleep.

‘This could run locally on the processors contained in the Alexa,’ Dr Gollakota said.

‘It’s running in real time, so you don’t need to store anything or send anything to the cloud.’

The researchers stress, however, more research is required to improve the tool’s accuracy. 

‘Right now, this is a good proof of concept using the 911 calls in the Seattle metropolitan area,’ Dr Gollakota said.

‘But we need to get access to more 911 calls related to cardiac arrest so that we can improve the accuracy of the algorithm further and ensure it generalises across a larger population.’

The researchers plan to one day commercialise this technology through Sound Life Sciences, a University of Washington spin-out.