Get smarter, faster. Subscribe to our daily newsletter.
Artificial intelligence model detects asymptomatic Covid-19 infections through cellphone-recorded coughs
The researchers trained the model on tens of thousands of samples of coughs, as well as spoken words.
Asymptomatic people who are infected with Covid-19 exhibit, by definition, no discernible physical symptoms of the disease. They are thus less likely to seek out testing for the virus, and could unknowingly spread the infection to others.
But it seems those who are asymptomatic may not be entirely free of changes wrought by the virus. MIT researchers have now found that people who are asymptomatic may differ from healthy individuals in the way that they cough. These differences are not decipherable to the human ear. But it turns out that they can be picked up by artificial intelligence.
In a paper published recently in the IEEE Journal of Engineering in Medicine and Biology, the team reports on an AI model that distinguishes asymptomatic people from healthy individuals through forced-cough recordings, which people voluntarily submitted through web browsers and devices such as cellphones and laptops.
The researchers trained the model on tens of thousands of samples of coughs, as well as spoken words. When they fed the model new cough recordings, it accurately identified 98.5 percent of coughs from people who were confirmed to have Covid-19, including 100 percent of coughs from asymptomatics — who reported they did not have symptoms but had tested positive for the virus.
The team is working on incorporating the model into a user-friendly app, which if FDA-approved and adopted on a large scale could potentially be a free, convenient, noninvasive prescreening tool to identify people who are likely to be asymptomatic for Covid-19. A user could log in daily, cough into their phone, and instantly get information on whether they might be infected and therefore should confirm with a formal test.
"The effective implementation of this group diagnostic tool could diminish the spread of the pandemic if everyone uses it before going to a classroom, a factory, or a restaurant," says co-author Brian Subirana, a research scientist in MIT's Auto-ID Laboratory.
Subirana's co-authors are Jordi Laguarta and Ferran Hueto, of MIT's Auto-ID Laboratory.
Prior to the pandemic's onset, research groups already had been training algorithms on cellphone recordings of coughs to accurately diagnose conditions such as pneumonia and asthma. In similar fashion, the MIT team was developing AI models to analyze forced-cough recordings to see if they could detect signs of Alzheimer's, a disease associated with not only memory decline but also neuromuscular degradation such as weakened vocal cords.
They first trained a general machine-learning algorithm, or neural network, known as ResNet50, to discriminate sounds associated with different degrees of vocal cord strength. Studies have shown that the quality of the sound "mmmm" can be an indication of how weak or strong a person's vocal cords are. Subirana trained the neural network on an audiobook dataset with more than 1,000 hours of speech, to pick out the word "them" from other words like "the" and "then."
The team trained a second neural network to distinguish emotional states evident in speech, because Alzheimer's patients — and people with neurological decline more generally — have been shown to display certain sentiments such as frustration, or having a flat affect, more frequently than they express happiness or calm. The researchers developed a sentiment speech classifier model by training it on a large dataset of actors intonating emotional states, such as neutral, calm, happy, and sad.
The researchers then trained a third neural network on a database of coughs in order to discern changes in lung and respiratory performance.
Finally, the team combined all three models, and overlaid an algorithm to detect muscular degradation. The algorithm does so by essentially simulating an audio mask, or layer of noise, and distinguishing strong coughs — those that can be heard over the noise — over weaker ones.
With their new AI framework, the team fed in audio recordings, including of Alzheimer's patients, and found it could identify the Alzheimer's samples better than existing models. The results showed that, together, vocal cord strength, sentiment, lung and respiratory performance, and muscular degradation were effective biomarkers for diagnosing the disease.
When the coronavirus pandemic began to unfold, Subirana wondered whether their AI framework for Alzheimer's might also work for diagnosing Covid-19, as there was growing evidence that infected patients experienced some similar neurological symptoms such as temporary neuromuscular impairment.
"The sounds of talking and coughing are both influenced by the vocal cords and surrounding organs. This means that when you talk, part of your talking is like coughing, and vice versa. It also means that things we easily derive from fluent speech, AI can pick up simply from coughs, including things like the person's gender, mother tongue, or even emotional state. There's in fact sentiment embedded in how you cough," Subirana says. "So we thought, why don't we try these Alzheimer's biomarkers [to see if they're relevant] for Covid."
"A striking similarity"
In April, the team set out to collect as many recordings of coughs as they could, including those from Covid-19 patients. They established a website where people can record a series of coughs, through a cellphone or other web-enabled device. Participants also fill out a survey of symptoms they are experiencing, whether or not they have Covid-19, and whether they were diagnosed through an official test, by a doctor's assessment of their symptoms, or if they self-diagnosed. They also can note their gender, geographical location, and native language.
To date, the researchers have collected more than 70,000 recordings, each containing several coughs, amounting to some 200,000 forced-cough audio samples, which Subirana says is "the largest research cough dataset that we know of." Around 2,500 recordings were submitted by people who were confirmed to have Covid-19, including those who were asymptomatic.
The team used the 2,500 Covid-associated recordings, along with 2,500 more recordings that they randomly selected from the collection to balance the dataset. They used 4,000 of these samples to train the AI model. The remaining 1,000 recordings were then fed into the model to see if it could accurately discern coughs from Covid patients versus healthy individuals.
Surprisingly, as the researchers write in their paper, their efforts have revealed "a striking similarity between Alzheimer's and Covid discrimination."
Without much tweaking within the AI framework originally meant for Alzheimer's, they found it was able to pick up patterns in the four biomarkers — vocal cord strength, sentiment, lung and respiratory performance, and muscular degradation — that are specific to Covid-19. The model identified 98.5 percent of coughs from people confirmed with Covid-19, and of those, it accurately detected all of the asymptomatic coughs.
"We think this shows that the way you produce sound, changes when you have Covid, even if you're asymptomatic," Subirana says.
The AI model, Subirana stresses, is not meant to diagnose symptomatic people, as far as whether their symptoms are due to Covid-19 or other conditions like flu or asthma. The tool's strength lies in its ability to discern asymptomatic coughs from healthy coughs.
The team is working with a company to develop a free pre-screening app based on their AI model. They are also partnerning with several hospitals around the world to collect a larger, more diverse set of cough recordings, which will help to train and strengthen the model's accuracy.
As they propose in their paper, "Pandemics could be a thing of the past if pre-screening tools are always on in the background and constantly improved."
Ultimately, they envision that audio AI models like the one they've developed may be incorporated into smart speakers and other listening devices so that people can conveniently get an initial assessment of their disease risk, perhaps on a daily basis.
This research was supported, in part, by Takeda Pharmaceutical Company Limited.
- Do you have a cold, the flu, or COVID-19? - Big Think ›
- Signs of Covid-19 may be hidden in speech signals | MIT News ... ›
Northwell Health is using insights from website traffic to forecast COVID-19 hospitalizations two weeks in the future.
- The machine-learning algorithm works by analyzing the online behavior of visitors to the Northwell Health website and comparing that data to future COVID-19 hospitalizations.
- The tool, which uses anonymized data, has so far predicted hospitalizations with an accuracy rate of 80 percent.
- Machine-learning tools are helping health-care professionals worldwide better constrain and treat COVID-19.
The value of forecasting<img type="lazy-image" data-runner-src="https://assets.rebelmouse.io/eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpbWFnZSI6Imh0dHBzOi8vYXNzZXRzLnJibC5tcy8yNTA0Njk2OC9vcmlnaW4uanBnIiwiZXhwaXJlc19hdCI6MTYyMzM2NDQzOH0.rid9regiDaKczCCKBsu7wrHkNQ64Vz_XcOEZIzAhzgM/img.jpg?width=980" id="2bb93" class="rm-shortcode" data-rm-shortcode-id="31345afbdf2bd408fd3e9f31520c445a" data-rm-shortcode-name="rebelmouse-image" data-width="1546" data-height="1056" />
Northwell emergency departments use the dashboard to monitor in real time.
Credit: Northwell Health<p>One unique benefit of forecasting COVID-19 hospitalizations is that it allows health systems to better prepare, manage and allocate resources. For example, if the tool forecasted a surge in COVID-19 hospitalizations in two weeks, Northwell Health could begin:</p><ul><li>Making space for an influx of patients</li><li>Moving personal protective equipment to where it's most needed</li><li>Strategically allocating staff during the predicted surge</li><li>Increasing the number of tests offered to asymptomatic patients</li></ul><p>The health-care field is increasingly using machine learning. It's already helping doctors develop <a href="https://care.diabetesjournals.org/content/early/2020/06/09/dc19-1870" target="_blank">personalized care plans for diabetes patients</a>, improving cancer screening techniques, and enabling mental health professionals to better predict which patients are at <a href="https://healthitanalytics.com/news/ehr-data-fuels-accurate-predictive-analytics-for-suicide-risk" target="_blank" rel="noopener noreferrer">elevated risk of suicide</a>, to name a few applications.</p><p>Health systems around the world have already begun exploring how <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7315944/" target="_blank" rel="noopener noreferrer">machine learning can help battle the pandemic</a>, including better COVID-19 screening, diagnosis, contact tracing, and drug and vaccine development.</p><p>Cruzen said these kinds of tools represent a shift in how health systems can tackle a wide variety of problems.</p><p>"Health care has always used the past to predict the future, but not in this mathematical way," Cruzen said. "I think [Northwell Health's new predictive tool] really is a great first example of how we should be attacking a lot of things as we go forward."</p>
Making machine-learning tools openly accessible<p>Northwell Health has made its predictive tool <a href="https://github.com/northwell-health/covid-web-data-predictor" target="_blank">available for free</a> to any health system that wishes to utilize it.</p><p>"COVID is everybody's problem, and I think developing tools that can be used to help others is sort of why people go into health care," Dr. Cruzen said. "It was really consistent with our mission."</p><p>Open collaboration is something the world's governments and health systems should be striving for during the pandemic, said Michael Dowling, Northwell Health's president and CEO.</p><p>"Whenever you develop anything and somebody else gets it, they improve it and they continue to make it better," Dowling said. "As a country, we lack data. I believe very, very strongly that we should have been and should be now working with other countries, including China, including the European Union, including England and others to figure out how to develop a health surveillance system so you can anticipate way in advance when these things are going to occur."</p><p>In all, Northwell Health has treated more than 112,000 COVID patients. During the pandemic, Dowling said he's seen an outpouring of goodwill, collaboration, and sacrifice from the community and the tens of thousands of staff who work across Northwell.</p><p>"COVID has changed our perspective on everything—and not just those of us in health care, because it has disrupted everybody's life," Dowling said. "It has demonstrated the value of community, how we help one another."</p>
Scientists used CT scanning and 3D-printing technology to re-create the voice of Nesyamun, an ancient Egyptian priest.
- Scientists printed a 3D replica of the vocal tract of Nesyamun, an Egyptian priest whose mummified corpse has been on display in the UK for two centuries.
- With the help of an electronic device, the reproduced voice is able to "speak" a vowel noise.
- The team behind the "Voices of the Past" project suggest reproducing ancient voices could make museum experiences more dynamic.
Howard et al.<p style="margin-left: 20px;">"While this approach has wide implications for heritage management/museum display, its relevance conforms exactly to the ancient Egyptians' fundamental belief that 'to speak the name of the dead is to make them live again'," they wrote in a <a href="https://www.nature.com/articles/s41598-019-56316-y#Fig3" target="_blank">paper</a> published in Nature Scientific Reports. "Given Nesyamun's stated desire to have his voice heard in the afterlife in order to live forever, the fulfilment of his beliefs through the synthesis of his vocal function allows us to make direct contact with ancient Egypt by listening to a sound from a vocal tract that has not been heard for over 3000 years, preserved through mummification and now restored through this new technique."</p>
Connecting modern people with history<p>It's not the first time scientists have "re-created" an ancient human's voice. In 2016, for example, Italian researchers used software to <a href="https://www.smithsonianmag.com/smart-news/hear-recreated-voice-otzi-iceman-180960570/" target="_blank">reconstruct the voice of Ötzi,</a> an iceman who was discovered in 1991 and is thought to have died more than 5,000 years ago. But the "Voices of the Past" project is different, the researchers note, because Nesyamun's mummified corpse is especially well preserved.</p><p style="margin-left: 20px;">"It was particularly suited, given its age and preservation [of its soft tissues], which is unusual," Howard told <em><a href="https://www.livescience.com/amp/ancient-egypt-mummy-voice-reconstructed.html" target="_blank">Live Science</a>.</em></p><p>As to whether Nesyamun's reconstructed voice will ever be able to speak complete sentences, Howard told <em><a href="https://abcnews.go.com/Weird/wireStory/ancient-voice-scientists-recreate-sound-egyptian-mummy-68482015" target="_blank">The Associated Press</a>, </em>that it's "something that is being worked on, so it will be possible one day."</p><p>John Schofield, an archaeologist at the University of York, said that reproducing voices from history can make museum experiences "more multidimensional."</p><p style="margin-left: 20px;">"There is nothing more personal than someone's voice," he told <em>The Associated Press.</em> "So we think that hearing a voice from so long ago will be an unforgettable experience, making heritage places like Karnak, Nesyamun's temple, come alive."</p>
A new study proposes mysterious axions may be found in X-rays coming from a cluster of neutron stars.
Are Axions Dark Matter?<span style="display:block;position:relative;padding-top:56.25%;" class="rm-shortcode" data-rm-shortcode-id="5e35ce24a5b17102bfce5ae6aecc7c14"><iframe type="lazy-iframe" data-runner-src="https://www.youtube.com/embed/e7yXqF32Yvw?rel=0" width="100%" height="auto" frameborder="0" scrolling="no" style="position:absolute;top:0;left:0;width:100%;height:100%;"></iframe></span>
New research suggests you can't fake your emotional state to improve your work life — you have to feel it.
What is deep acting?<img type="lazy-image" data-runner-src="https://assets.rebelmouse.io/eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJpbWFnZSI6Imh0dHBzOi8vYXNzZXRzLnJibC5tcy8yNTQ1NDk2OS9vcmlnaW4uanBnIiwiZXhwaXJlc19hdCI6MTYxNTY5MzA0Nn0._s7aP25Es1CInq51pbzGrUj3GtOIRWBHZxCBFnbyXY8/img.jpg?width=1245&coordinates=333%2C-1%2C333%2C-1&height=700" id="ddf09" class="rm-shortcode" data-rm-shortcode-id="9dc42c4d6a8e372ad7b72907b46ecd3f" data-rm-shortcode-name="rebelmouse-image" data-width="1245" data-height="700" />
Arlie Russell Hochschild (pictured) laid out the concept of emotional labor in her 1983 book, "The Managed Heart."
Credit: Wikimedia Commons<p>Deep and surface acting are the principal components of emotional labor, a buzz phrase you have likely seen flitting about the Twittersphere. Today, "<a href="https://www.bbc.co.uk/bbcthree/article/5ea9f140-f722-4214-bb57-8b84f9418a7e" target="_blank">emotional labor</a>" has been adopted by groups as diverse as family counselors, academic feminists, and corporate CEOs, and each has redefined it with a patented spin. But while the phrase has splintered into a smorgasbord of pop-psychological arguments, its initial usage was more specific.</p><p>First coined by sociologist Arlie Russell Hochschild in her 1983 book, "<a href="https://www.ucpress.edu/book/9780520272941/the-managed-heart" target="_blank">The Managed Heart</a>," emotional labor describes the work we do to regulate our emotions on the job. Hochschild's go-to example is the flight attendant, who is tasked with being "nicer than natural" to enhance the customer experience. While at work, flight attendants are expected to smile and be exceedingly helpful even if they are wrestling with personal issues, the passengers are rude, and that one kid just upchucked down the center aisle. Hochschild's counterpart to the flight attendant is the bill collector, who must instead be "nastier than natural."</p><p>Such personas may serve an organization's mission or commercial interests, but if they cause emotional dissonance, they can potentially lead to high emotional costs for the employee—bringing us back to deep and surface acting.</p><p>Deep acting is the process by which people modify their emotions to match their expected role. Deep actors still encounter the negative emotions, but they devise ways to <a href="http://www.selfinjury.bctr.cornell.edu/perch/resources/what-is-emotion-regulationsinfo-brief.pdf" target="_blank">regulate those emotions</a> and return to the desired state. Flight attendants may modify their internal state by talking through harsh emotions (say, with a coworker), focusing on life's benefits (next stop Paris!), physically expressing their desired emotion (smiling and deep breaths), or recontextualizing an inauspicious situation (not the kid's fault he got sick).</p><p>Conversely, surface acting occurs when employees display ersatz emotions to match those expected by their role. These actors are the waiters who smile despite being crushed by the stress of a dinner rush. They are the CEOs who wear a confident swagger despite feelings of inauthenticity. And they are the bouncers who must maintain a steely edge despite humming show tunes in their heart of hearts.</p><p>As we'll see in the research, surface acting can degrade our mental well-being. This deterioration can be especially true of people who must contend with negative emotions or situations inside while displaying an elated mood outside. Hochschild argues such emotional labor can lead to exhaustion and self-estrangement—that is, surface actors erect a bulwark against anger, fear, and stress, but that disconnect estranges them from the emotions that allow them to connect with others and live fulfilling lives.</p>
Don't fake it till you make it<p>Most studies on emotional labor have focused on customer service for the obvious reason that such jobs prescribe emotional states—service with a smile or, if you're in the bouncing business, a scowl. But <a href="https://eller.arizona.edu/people/allison-s-gabriel" target="_blank">Allison Gabriel</a>, associate professor of management and organizations at the University of Arizona's Eller College of Management, wanted to explore how employees used emotional labor strategies in their intra-office interactions and which strategies proved most beneficial.</p><p>"What we wanted to know is whether people choose to engage in emotion regulation when interacting with their co-workers, why they choose to regulate their emotions if there is no formal rule requiring them to do so, and what benefits, if any, they get out of this effort," Gabriel said in <a href="https://www.sciencedaily.com/releases/2020/01/200117162703.htm" target="_blank">a press release</a>.</p><p>Across three studies, she and her colleagues surveyed more than 2,500 full-time employees on their emotional regulation with coworkers. The survey asked participants to agree or disagree with statements such as "I try to experience the emotions that I show to my coworkers" or "I fake a good mood when interacting with my coworkers." Other statements gauged the outcomes of such strategies—for example, "I feel emotionally drained at work." Participants were drawn from industries as varied as education, engineering, and financial services.</p><p>The results, <a href="https://psycnet.apa.org/doiLanding?doi=10.1037%2Fapl0000473" target="_blank" rel="noopener noreferrer">published in the Journal of Applied Psychology</a>, revealed four different emotional strategies. "Deep actors" engaged in high levels of deep acting; "low actors" leaned more heavily on surface acting. Meanwhile, "non-actors" engaged in negligible amounts of emotional labor, while "regulators" switched between both. The survey also revealed two drivers for such strategies: prosocial and impression management motives. The former aimed to cultivate positive relationships, the latter to present a positive front.</p><p>The researchers found deep actors were driven by prosocial motives and enjoyed advantages from their strategy of choice. These actors reported lower levels of fatigue, fewer feelings of inauthenticity, improved coworker trust, and advanced progress toward career goals. </p><p>As Gabriel told <a href="https://www.psypost.org/2021/01/new-psychology-research-suggests-deep-acting-can-reduce-fatigue-and-improve-your-work-life-59081" target="_blank" rel="noopener noreferrer">PsyPost in an interview</a>: "So, it's a win-win-win in terms of feeling good, performing well, and having positive coworker interactions."</p><p>Non-actors did not report the emotional exhaustion of their low-actor peers, but they also didn't enjoy the social gains of the deep actors. Finally, the regulators showed that the flip-flopping between surface and deep acting drained emotional reserves and strained office relationships.</p><p>"I think the 'fake it until you make it' idea suggests a survival tactic at work," Gabriel noted. "Maybe plastering on a smile to simply get out of an interaction is easier in the short run, but long term, it will undermine efforts to improve your health and the relationships you have at work. </p><p>"It all boils down to, 'Let's be nice to each other.' Not only will people feel better, but people's performance and social relationships can also improve."</p>
You'll be glad ya' decided to smile<span style="display:block;position:relative;padding-top:56.25%;" class="rm-shortcode" data-rm-shortcode-id="88a0a6a8d1c1abfcf7b1aca8e71247c6"><iframe type="lazy-iframe" data-runner-src="https://www.youtube.com/embed/QOSgpq9EGSw?rel=0" width="100%" height="auto" frameborder="0" scrolling="no" style="position:absolute;top:0;left:0;width:100%;height:100%;"></iframe></span><p>But as with any research that relies on self-reported data, there are confounders here to untangle. Even during anonymous studies, participants may select socially acceptable answers over honest ones. They may further interpret their goal progress and coworker interactions more favorably than is accurate. And certain work conditions may not produce the same effects, such as toxic work environments or those that require employees to project negative emotions.</p><p>There also remains the question of the causal mechanism. If surface acting—or switching between surface and deep acting—is more mentally taxing than genuinely feeling an emotion, then what physiological process causes this fatigue? <a href="https://www.frontiersin.org/articles/10.3389/fnhum.2019.00151/full" target="_blank">One study published in the <em>Frontiers in Human Neuroscience</em></a><em> </em>measured hemoglobin density in participants' brains using an fNIRS while they expressed emotions facially. The researchers found no significant difference in energy consumed in the prefrontal cortex by those asked to deep act or surface act (though, this study too is limited by a lack of real-life task).<br></p><p>With that said, Gabriel's studies reinforce much of the current research on emotional labor. <a href="https://journals.sagepub.com/doi/abs/10.1177/2041386611417746" target="_blank">A 2011 meta-analysis</a> found that "discordant emotional labor states" (read: surface acting) were associated with harmful effects on well-being and performance. The analysis found no such consequences for deep acting. <a href="https://doi.apa.org/doiLanding?doi=10.1037%2Fa0022876" target="_blank" rel="noopener noreferrer">Another meta-analysis</a> found an association between surface acting and impaired well-being, job attitudes, and performance outcomes. Conversely, deep acting was associated with improved emotional performance.</p><p>So, although there's still much to learn on the emotional labor front, it seems Van Dyke's advice to a Leigh was half correct. We should put on a happy face, but it will <a href="https://bigthink.com/design-for-good/everything-you-should-know-about-happiness-in-one-infographic" target="_self">only help if we can feel it</a>.</p>
Archaeologists discover a cave painting of a wild pig that is now the world's oldest dated work of representational art.