A Health check App sensitive to asymptomatic COVID may be in the works

[.Scott]

TL;DR Summary

Promising results from an algorithm that checks for asymptomatic COVID (and other conditions) based on the sound of a forced cough.

https://www.embs.org/ojemb/articles...igence-diagnosis-using-only-cough-recordings/

A study has been posted in Engineering in Medicine and Biology (article) that describes the methods and results from an MIT algorithm as applied to asymptomatic COVID-19. Per the article: "For asymptomatic subjects it achieves sensitivity of 100% with a specificity of 83.2%."

Here's a report on that article.

 

[berkeman]

Summary:: Promising results from an algorithm that checks for asymptomatic COVID (and other conditions) based on the sound of a forced cough.

https://www.embs.org/ojemb/articles...igence-diagnosis-using-only-cough-recordings/

A study has been posted in Engineering in Medicine and Biology (article) that describes the methods and results from an MIT algorithm as applied to asymptomatic COVID-19. Per the article: "For asymptomatic subjects it achieves sensitivity of 100% with a specificity of 83.2%."

Here's a report on that article.

Wow, impressive. Two questions:

** Does it work if there is an N95 mask held between the patient and my cellphone?

** Is it available in the Google Store?

 

[mfb]

100% just means their sample size was too small. But let's use that number:

For asymptomatic subjects it achieves sensitivity of 100% with a specificity of 83.2%.

If you have 100 people using that app you might have 0-5 (depending on the location) who actually have the disease, they will show up as positive, and you'll also get 17 who will show up as positive without having it. In other words: Most positive results will be false positives. If the whole population downloads that and tests itself twice per week then you'll want to re-test 5% of the population every day. UAE and Denmark are leading at testing 1% of their population every day at the moment, the US is testing 0.3%, France is testing 0.4%.
You could ask everyone to self-isolate for a week whenever they test positive, but people won't do that with a 17% risk to get a false positive result.