OT: MIT AI detects Covid cough...

[Martin Brown]

MIT have announced AI based screening software that they claim can
recognise Covid cough with 98.5% accuracy even in people who are
asymptomatic. If it is true then this could be a great help in stopping
people who are infected from spreading the virus to others.

Very much faster than even the fastest biochemical test and potentially
something that an app on every smartphone could do.

https://www.theregister.com/2020/10/31/ai_covid_cough/

original IEEE paper is here:

https://ieeexplore.ieee.org/document/9208795

(and free access)

It is about time we had some good news on this pandemic.
England goes back into lockdown this Thursday

At least we aren\'t doing quite as badly as Belgium or Switzerland which
are now in almost perfect textbook exponential growth!

--
Regards,
Martin Brown

 

[Jeroen Belleman]

On 2020-11-02 17:32, Martin Brown wrote:

MIT have announced AI based screening software that they claim can
recognise Covid cough [...]


It is about time we had some good news on this pandemic.[...]

And this is good news. Good grief. AI to the rescue, I feel
better already.

THIS IS NONSENSE!

Jeroen Belleman

 

[Jeff Liebermann]

On Mon, 2 Nov 2020 16:32:20 +0000, Martin Brown
<\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

MIT have announced AI based screening software that they claim can
recognise Covid cough with 98.5% accuracy even in people who are
asymptomatic. If it is true then this could be a great help in stopping
people who are infected from spreading the virus to others.

Very much faster than even the fastest biochemical test and potentially
something that an app on every smartphone could do.

https://www.theregister.com/2020/10/31/ai_covid_cough/

original IEEE paper is here:

https://ieeexplore.ieee.org/document/9208795

I\'m not sure that having someone cough into my smartphone is a good
idea, especially if I need to later make or receive a phone call.
Perhaps with an external microphone and a disinfectant tolerant
microphone. Also, I wonder if the test recordings were made with or
without a face mask. Keyword search of original PDF shows no relevant
mention of mask or facemask. Still, it sure would be nice if it works
as advertised.

Note that this was previously done in July 2020 at Cambridge:
<https://covid-19-sounds.org/en/>
Apps and web apps used to collect the sample recordings:
<https://play.google.com/store/apps/details?id=uk.ac.cam.cl.covid19sounds>
<https://apps.apple.com/gb/app/covid19-sounds/id1510614550>
<https://covid-19-sounds.org/en/app/>
I could not determine if there have been any results published or if
this is still an ongoing study.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558

 

[Martin Brown]

On 02/11/2020 22:58, Jeff Liebermann wrote:

On Mon, 2 Nov 2020 16:32:20 +0000, Martin Brown
\'\'\'newspam\'\'\'@nonad.co.uk> wrote:

MIT have announced AI based screening software that they claim can
recognise Covid cough with 98.5% accuracy even in people who are
asymptomatic. If it is true then this could be a great help in stopping
people who are infected from spreading the virus to others.

Very much faster than even the fastest biochemical test and potentially
something that an app on every smartphone could do.

https://www.theregister.com/2020/10/31/ai_covid_cough/

original IEEE paper is here:

https://ieeexplore.ieee.org/document/9208795

I\'m not sure that having someone cough into my smartphone is a good
idea, especially if I need to later make or receive a phone call.

I think that the idea is that they cough into their *own* smartphone
every morning before they go outside. If it works it would be magic!

https://covid-19-sounds.org/en/
Apps and web apps used to collect the sample recordings:
https://play.google.com/store/apps/details?id=uk.ac.cam.cl.covid19sounds
https://apps.apple.com/gb/app/covid19-sounds/id1510614550
https://covid-19-sounds.org/en/app/
I could not determine if there have been any results published or if
this is still an ongoing study.

Published although they make less grandiose claims than MIT.

https://dl.acm.org/doi/10.1145/3394486.3412865

Looks like there might be scope for something *very* useful here.

After all the gold standard PCR test gives false negatives about 20% of
the time even when testing people with all of the symptoms of Covid!

The biochemical test swab is also somewhat unpleasant to obtain.

--
Regards,
Martin Brown

 

[Martin Brown]

On 02/11/2020 18:04, Jeroen Belleman wrote:

On 2020-11-02 17:32, Martin Brown wrote:
MIT have announced AI based screening software that they claim can
recognise Covid cough [...]


It is about time we had some good news on this pandemic.[...]

And this is good news. Good grief. AI to the rescue, I feel
better already.

THIS IS NONSENSE!

It might not be - in which case it will depend on how quickly they can
turn it into an app that will work on most phones. I don\'t really care
if the algorithm internals are understood or not so long as it works as
advertised. A 98% reliable quick test would be a game changer.
(anything better than 80% would be too)

I grant you that MIT has previous for dodgy press releases.

Cambridge UK is also working on something similar.

--
Regards,
Martin Brown

 

[Tom Gardner]

On 05/11/20 11:06, Martin Brown wrote:

I was tested last month as a part of IPSOS MORI population survey. If you do it
right the swab is on the edge of causing the gag reflex. The nose sample is an
even weirder sensation to take.

I had an /antigen/ test last month, since I\'m a \"Father of the 90s\"
(i.e. my daughter is in the ALSPAC Choldren of the 90s cohort).

It is a mere pinprick but the result has to be assessed between
5mins and 10mins after taking the sample.

As expected we were both negative to both antigens.

 

[Martin Brown]

On 05/11/2020 16:38, Tom Gardner wrote:

On 05/11/20 11:06, Martin Brown wrote:

I was tested last month as a part of IPSOS MORI population survey. If
you do it right the swab is on the edge of causing the gag reflex. The
nose sample is an even weirder sensation to take.

I had an /antigen/ test last month, since I\'m a \"Father of the 90s\"
(i.e. my daughter is in the ALSPAC Choldren of the 90s cohort).

It is a mere pinprick but the result has to be assessed between
5mins and 10mins after taking the sample.

As expected we were both negative to both antigens.

My friend who caught it returning from an expedition in Peru through
London at the height of the first wave in March is now negative for
Covid antibodies again back to baseline within 6 months. He was part of
the plasma donor scheme until he stopped having any antibodies to give.

--
Regards,
Martin Brown