I can't 'prove' it since the reach of the subject here is wider than any scientific experiment can reasonably capture.
I can give you something to think about, though:
We often categorize mental illness based on behaviors, but certain behaviors can be caused by a lot of different things.
In traditional medicine, we can typically find a root cause eventually. Your kidney hurts because there's a nail in it.
But without the technology to fully map the brain, we're missing that. What if, for example, you were born with brain damage that made it very hard for you to socialize, but through great effort you learned how anyway? You definitely have some kind of disorder, but it would never be diagnosed because your behavior doesn't reflect it anymore.
In contrast, even if you stopped feeling pain from the nail in your kidney, a scan would tell us for certain it was still there.
As a whole, a person's behavior and their mind are immensely complicated to a degree we lack the ability to fully comprehend, and to put certain behavioral patterns into certain boxes is only useful for the sake of quickly conveying them to others, but if presented as the 'cause' of symptoms it falls short of ever proving that.