I think the general point of decision is still about "personally identifiable information".

So my (IANAL) understanding is: If the data is somehow tied to your real life identity, it's protected. E.g., the headphone brand might be protected because it's uniquely associated with your user account which is uniquely tied to a real-life identity.

I guess, whether or not the facial recognition output is protected, might depend on whether you want to use it to identify the person behind it - e.g., if it's an identifier to a database, it might be protected. If it's an emotion score or eye tracking analysis, it might not be protected, unless it's associated with your identity by some other process.