Not sure why you're so hot on cough only, but you can calculate the confidence intervals yourself quite easily using this nifty tool:
https://surveysystem.com/sscalc.htm
For this, use the second calculator. I'll work an example with you for pneumonia on cough-only:
Sample size is 123.
Population is... everybody in the world, I used a million to represent "infinite"
Results say PPA is 80%.
Click Calculate
Get 7.07
What that says is that if you choose any random 123 people with pneumonia then you can be 95% confident that your algorithm will correctly diagnose between 72.93% and 87.07% of them as having the disease.
View attachment 689808