My wife has been following suspected COVID patients and has said since the beggining their cough is different than normal. Makes sense thay ML would work is this scenario.

This anecdote actually makes me a lot more confident the tool might work. The article sounded to me like they just threw something at the wall, and it's not that hard to get a model that looks good on paper when you take the liberties they did.

But if this study serves as a PoC to back up that real-world observation, then this is quite a promising approach!

I was coming off a cold from February when I got it in March. My latent cough was no different for the four days I had unusual symptoms.

The article claims that the difference is indecipherable.

What is the false positive rate, though?

Like 20% - but that doesn’t matter as much in terms of ensure no sick people go out in public.

While people can be fairly dumb they may be able to cope with a warning that a cough listening device is not very accurate?