Free AI web copilot to create summaries, insights and extended knowledge, download it at here

Abstract

rchers were already investigating AI models and biomarkers for Alzheimer's disease when COVID-19 broke out. They found that the <a href="https://healthitanalytics.com/news/artificial-intelligence-models-identify-alzheimers-cognitive-decline?__cf_chl_captcha_tk__=1b0fe7c0b1869a4d93417c6cbf62d2ff51311b6b-1604828800-0-Aajh8ukCb39FCwHyfPjpVGmo-KPdhnC3kMrilq316q1pqI-lt83t24aMSMHo8mlKpQfPAmw20ksKgsQupoehpbys96s3l44H2t-gYvlcO6YShBEw4Lxdh8sN_ojy7uKv6KZadTY3XuuECjNzucAaKkQ00ZZyOcGpPQo1A3otff8W0Mh5FfqkgMDerYYdfyBLUOBFHUIsTobLlIkqJsa0DD5MLWavQk5Z8ifXkW4_F6uAALnE_U7S78IzJH4OReEJa1DWcl7j3yAglkvttFNqypb8vumlswoiK2o8wnAazXfl25r8Bgc4opQr6jeaM1I6_Cfkuy2CgvlUOT8SICpASmfZPHPe-5QIKzZ1vWxjdX50sipB4TJamHLw9v6cafobUzhbqCTl9fXMHq8NPFxhMH9K8k5h6HJhsVxVKHjO_07D0VwN8B2CGwHFp6p5WkoQXiIrRDf3r82WjUYr36UslkVuq9nyFHCfV_Bi3taDbgzi2HU3TkyK_lAGMnGGvE9lJGCqQBJcICkpeJ-Hj2VzFuUQZdzjGt9XMoDP1_6s8egCXmMPY7wGw3pg6owGpzxEH7IhWsVC0cm8HCvgs0gILPg">AI model </a>was able to detect a rapid cognitive decline long before the actual clinical symptoms would even be apparent. This was helping them to develop drugs that could stave off this disease and help people live longer and more active lives.</p><p id="d7ca">The four biomarkers they were using for the Alzheimers AI model were</p><ul><li>Change in vocal cords</li><li>Muscular degeneration</li><li>Emotional analysis</li><li>Respiratory tract and lung changes</li></ul><p id="e532">They found that these changes also were associated with lung and respiratory diseases such as COVID-19 so they were able to adopt the model accordingly.</p><p id="e7cc">The way a person coughs can indicate an emotional state and even a person’s gender and nationality. There was also a link with <a href="https://www.health.harvard.edu/blog/the-hidden-long-term-cognitive-effects-of-covid-2020100821133">cognitive decline</a> as both diseases manifest some of these symptoms. There were other similarities in these diseases as regards voice muscle weakness and also a decline in memory ability. This explains why the Alzheimers AI model was such a sound base.</p><p id="4be9"><b><i>So far their data and research have had an impressive accuracy rate of between 97% and 100%.</i></b></p><h2 id="f68f">How were samples of coughs gathered?</h2><p id="dfdf">Researchers collected over 70,000 recordings of volunteers’ forced or spontaneous coughs which were sent via laptops, computers, and mobile phones. They were asked to fill in a questionnaire about their COVID -19 status. Then 2,600 recordings from COVID patients were used as part of the 4,000 total to train the AI model. They are working to expand their data set but they have gathered all their data from multiple ethnic groups and nationalities to avoid any potential bias.</p><p id="5343">Similar

Options

research has been carried out by the <a href="https://www.covid-19-sounds.org/en/">University of Cambridge</a> in the UK and the European Research Council They were able to gather 30,000 recordings obtained from about 400 members of the general public. This research has had an 80% success rate.</p><p id="1072">The <a href="https://www.biometricupdate.com/202004/cmu-researcher-says-voice-recognition-can-spot-covid-19-cough">Carnegie Mellon University (PA)</a> has developed a <a href="https://cvd.lti.cmu.edu/">COVID-19 voice detector</a> app for mobile phones but this has not yet been approved by the FDA. However, they are researching the use of an algorithm in detecting the COVID-19 cough and also the fever associated with this condition.</p><p id="e75b">NOVOIC is a research company in the UK that is collaborating with Oxford and Cambridge universities in the UK. It is doing similar research and has asked for volunteers to donate their coughs which they can do on their <a href="https://www.novoic.com/covid-19">site here.</a></p><h2 id="1e41">Doubts about the research</h2><p id="3906">Some researchers and commentators have remarked that forced coughs may not be the same as a natural cough and wonder whether this would skew the data. It may well be a <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5394006/#:~:text=The%20term%20psychogenic%20cough%20has,a%20psychiatric%20or%20psychological%20basis.">psychogenic cough</a> which will not help. Also, there do not seem to be instructions on how exactly to use the forced cough on the donating coughs websites that I have observed.</p><p id="9c4d">Nurses have often observed that in mild COVID-19 cases, the cough is not one of the symptoms so they wonder how effective this research can be.</p><h2 id="b259">Clinical trials</h2><p id="8c5b">The MIT team is working with four hospitals around the world in order to validate these models in a clinical setting. The next step will be to develop an app that can give instant feedback to anybody with a mobile phone.</p><p id="c177">This mass screening will be far more effective than the present track and trace system in place. It can alert possible COVID-19 positive cases to take a diagnostic test. It can be used for screening workers, students, and the general public. In this way, clusters of infection can be avoided or at least kept under close control.</p><h2 id="fbbd">Conclusion</h2><p id="88d8">The research so far is promising and the arrival of such a user-friendly app for COVID-19 is really attractive. This will be so much more comfortable and simpler than nasal swab or blood tests. It will be instantly available once it is approved by the FDA. Not to mention that there is no waiting for the results.</p></article></body>

Coronavirus

AI Model Used by MIT Researchers to Detect a COVID-19 Cough

How soon will an app be available?

While watching an Italian TV talk show last night, I listened to a biologist and journalist, Barbara Gallavotti, who was being interviewed. She spoke about how researchers at MIT were developing an app that could tell whether a person was an asymptomatic COVID-19 case or whether they were free of the virus simply by analyzing the sound of their coughs. Her initial reaction was that it must be some kind of joke!

She investigated its publication in the IEEXplore journal and discovered it was indeed promising. She then played a sample of two people coughing. One was asymptomatic while the other indicated that they were COVID-19 free and could go out without infecting anyone else.

I was unable to tell the difference between the two coughs, like most people on the show. But the researchers have done extensive research on this AI model. The prospects are promising that they may be able to develop a suitable phone app for this. This would replace the tracking and tracing apps which have been a flop in Germany, Italy, and Australia. The UK government is now under pressure to urgently change its test and trace system. The much-vaunted Operation Moonshot which was supposed to be groundbreaking missed 50% of COVID-19 positive cases in its pilot stage.

While we wait for a vaccine to be developed, this new app could be such a useful, quick, and easy screening tool. It would alert people to get tested and would reassure others that they were not carrying the virus.

How and why this research started

The researchers were already investigating AI models and biomarkers for Alzheimer's disease when COVID-19 broke out. They found that the AI model was able to detect a rapid cognitive decline long before the actual clinical symptoms would even be apparent. This was helping them to develop drugs that could stave off this disease and help people live longer and more active lives.

The four biomarkers they were using for the Alzheimers AI model were

Change in vocal cords
Muscular degeneration
Emotional analysis
Respiratory tract and lung changes

They found that these changes also were associated with lung and respiratory diseases such as COVID-19 so they were able to adopt the model accordingly.

The way a person coughs can indicate an emotional state and even a person’s gender and nationality. There was also a link with cognitive decline as both diseases manifest some of these symptoms. There were other similarities in these diseases as regards voice muscle weakness and also a decline in memory ability. This explains why the Alzheimers AI model was such a sound base.

So far their data and research have had an impressive accuracy rate of between 97% and 100%.

How were samples of coughs gathered?

Researchers collected over 70,000 recordings of volunteers’ forced or spontaneous coughs which were sent via laptops, computers, and mobile phones. They were asked to fill in a questionnaire about their COVID -19 status. Then 2,600 recordings from COVID patients were used as part of the 4,000 total to train the AI model. They are working to expand their data set but they have gathered all their data from multiple ethnic groups and nationalities to avoid any potential bias.

Similar research has been carried out by the University of Cambridge in the UK and the European Research Council They were able to gather 30,000 recordings obtained from about 400 members of the general public. This research has had an 80% success rate.

The Carnegie Mellon University (PA) has developed a COVID-19 voice detector app for mobile phones but this has not yet been approved by the FDA. However, they are researching the use of an algorithm in detecting the COVID-19 cough and also the fever associated with this condition.

NOVOIC is a research company in the UK that is collaborating with Oxford and Cambridge universities in the UK. It is doing similar research and has asked for volunteers to donate their coughs which they can do on their site here.

Doubts about the research

Some researchers and commentators have remarked that forced coughs may not be the same as a natural cough and wonder whether this would skew the data. It may well be a psychogenic cough which will not help. Also, there do not seem to be instructions on how exactly to use the forced cough on the donating coughs websites that I have observed.

Nurses have often observed that in mild COVID-19 cases, the cough is not one of the symptoms so they wonder how effective this research can be.

Clinical trials

The MIT team is working with four hospitals around the world in order to validate these models in a clinical setting. The next step will be to develop an app that can give instant feedback to anybody with a mobile phone.

This mass screening will be far more effective than the present track and trace system in place. It can alert possible COVID-19 positive cases to take a diagnostic test. It can be used for screening workers, students, and the general public. In this way, clusters of infection can be avoided or at least kept under close control.

Conclusion

The research so far is promising and the arrival of such a user-friendly app for COVID-19 is really attractive. This will be so much more comfortable and simpler than nasal swab or blood tests. It will be instantly available once it is approved by the FDA. Not to mention that there is no waiting for the results.