Well, you wouldn't. It's largely a metaphysical stance rather than a scientific one[0], which I suspect is precisely why modern medicine is struggling to engage with it.

[0] Mostly. Part of it falls under Dennett's notion of an intentional stance, but that requires us to take self-report seriously, and so precludes a number of experimental designs.

give patients a strong sedative (or after a period of sleep deprivation) and administer the drug while they're asleep?

Could give them a psychoactive that's not reputed to help with depression. As long as they don't know what a trip looks like they won't know you gave them DMT when they're talking to the machine elves.

You don’t. It’s fundamentally incompatible. But that’s fine.

People are able to acquire knowledge in ways besides randomized, controlled, double-blind trials. Nearly all knowledge is acquired in some other way. I say this as someone who writes A/B testing software for a living.