I think there are several questions around this particular situation. Like, is recording a medical session verbatim a good idea, ever? I know personally there are times when I ask my doctor (who I now know pretty well) not to put something in the record as we talk, for example. If recording is happening, I think patients should be able to it.
Does freeing the doctor from having to type everything mean they’re more attentive and can process more information… or is the act of slowing down for the doctor to live-summarize your conversation actually make them focus and listen more? I kinda lean towards the latter, but I don’t know if this has been studied.
But to the original question… I think it’s possible to do truly private transcription, and you pretty much described it: local only processing, remove the audio file when complete. I would add that the doctor should immediately review it and correct any mistakes while the conversation is fresh in his/her mind. And finally, I think the conversation should be summarized, not stored verbatim.
My problem is that I don’t trust my doctor or the person in the practice making buying choices to be able to vet the claims of the transcription device vendor. My doc is smart, but he’s no computer security wizard.
Maybe if we had specific privacy regulations around medical transcriptions and AI processing, of PHI, and then vendors were tested for compliance, then maybe.