People with asymptomatic COVID-19 can spread the disease without any outward signs that they are sick. But a newly developed AI, with a sharp algorithmic ear, can detect asymptomatic instances from people’s coughs, according to a new study.
A team of researchers at MIT recently developed one artificial intelligence The model can detect asymptomatic COVID-19 cases by listening for small differences in cough between healthy and infected people. Researchers are currently testing their AI in clinical trials and have begun the process of getting FDA permission for it to be used as a screening tool.
The algorithm is based on earlier models that the team developed to detect such conditions pneumonia, asthma and even AlzheimerA condition of dementia can also cause other deterioration in the body such as laryngeal laryngeal impairment and respiratory performance.
Related: Update directly to Coronavirus
Indeed, it is the Alzheimer̵7;s model that the researchers tuned in their efforts to detect COVID-19. Co-author Brian Subirana, a research scientist at MIT’s Auto-ID Laboratory: “The sound when speaking and coughing are both influenced by the vocal cords and surrounding organs. said in a statement. “Things we easily get from our ability to speak fluently, AI can recognize simply from coughs, including things like the person’s gender, mother tongue or even emotional state. Emotional reality is tied to the way you cough. ”
First, they created a website where volunteers – both healthy people and people with COVID-19 – could record coughs with their cell phones or computers; they also filled out a survey with questions about their diagnosis and any symptoms they were experiencing. People are asked to record “forced coughs”, such as the cough you make when your doctor tells you to cough when listening to your chest with a stethoscope.
Through this website, researchers collected more than 70,000 individual recordings of forced cough patterns, according to the statement. Of these, 2,660 were from patients with COVID-19, with or without symptoms. They then used 4,256 samples to train their AI model and 1,064 samples to test their model to see if it could detect a difference in coughing spells between patients with COVID-19 and humans. healthy or not.
They found that their AI could recognize a difference in coughing related to four specific features of COVID-19 (also used in their Alzheimer’s algorithm) – muscle deterioration, strength. vocal cords, emotions such as suspicion and disappointment and lung respiratory and performance.
The AI model accurately identified 98.5% of people with COVID-19 and accurately eliminated COVID-19 in 94.2% of those without. For asymptomatic individuals, the model accurately identified 100% of those with COVID-19 and accurately ruled out COVID-19 in 83.2% of those without.
Dr. Anthony Lubinsky, medical director of respiratory care at NYU Langone Tisch Hospital, who was not involved in the study, said this was “a pretty encouraging set of numbers” and the results were “very interesting. “.
But “whether this works well enough in a real world context to recommend its use as a screening tool will need more research,” Lubinsky told Live Science. Furthermore, more research is needed to ensure AI will accurately assess coughs of people of all ages and ethnicities, he said (The authors also mention this limitation in their article).
Related: The most promising vaccine candidates COVID-19
If the doctor listens for the forced coughing sound of a person with COVID-19 with no symptoms, they probably won’t be able to hear anything out of the ordinary. “It’s not something the human ear can easily do,” Lubinsky said. While further research is definitely needed, if the software proves to be effective, then this AI – which will have a linked application if approved – can be “very helpful” to find out. asymptomatic cases of COVID-19, especially if the tool is cheap and easy. to use, he added.
Fully AI can help limit the spread of pandemic Subirana told Live Science in an email by helping to detect people with the disease who have no symptoms. AI can also detect differences between people with other diseases like flu and those with COVID-19, but much better at distinguishing COVID-19 cases from healthy cases, he said.
The team is currently seeking regulatory approval for an AI model matching application, which could launch next month, he said. They are also testing their AI in clinical trials at several hospitals around the world, according to the article.
And they’re not the only group working on detecting COVID-19 through sound. Similar projects are underway at Cambridge University, Carnegie Mellon University and the UK startup Novoic, according to BBC.
“Pandemics could be a thing of the past if pre-screening tools are always enabled in the background and constantly improving,” the authors write in the article. Those always listening, they write, could be smart speakers or smartphones.
The study, partially supported by the Takeda Pharmaceutical Company Limited, was published September 30 in the journal. IEEE Open Journal of Engineering in Medicine and Biology.
Originally published on Live Science.