Hmm... perhaps I should have inserted (not-serious face) in the comment about cough-accent.
It's quite likely that the ResApp algorithm will be discounting or ignoring vocalization and articulation in the cough and focusing on lung-sounds; wheeze (indicating constricted airways), gurgle (indicating level and consistency of congestion), and so on.
Distinguishing between pulmonary and naso-pharyngeal sounds was probably a significant training phase with early data.
RAP Price at posting:
43.0¢ Sentiment: Buy Disclosure: Held
(20min delay)