Anecdotally, it caught a good 80-90%+ of instances where users touched their face. No, it's absolutely not perfect, but it was an app I built over a weekend and I never billed it as a medical app. On their request, I even removed all references to health. It was always meant to help nudge people towards awareness of touching their face and didn't need to be 100% accurate.

By the end (and the website doesn't show the latest screenshots since I gave up updating it after getting the run around for weeks), it was even so neutral it didn't make a claim that touching your face was even good or bad. You could have interpreted it as a trainer to touch your face more if you wanted.