How does an ai scribe actually handle multiple speakers in a room during patient visits?
I work in a small urgent care clinic and we often have situations where a patient's family member is also in the room talking, or a nurse pops in to ask a quick question mid-visit. I've been looking into ai scribe tools for a few weeks now but I genuinely cannot find a clear answer on how they handle that kind of overlapping audio.
Most of the marketing material just says things like 'captures the entire conversation' but that doesn't tell me much. Like, does the tool try to separate speakers? Does it get confused and start attributing the wrong statements to the provider versus the patient? Or does it just transcribe everything as one big block and leave you to sort it out?
I ask because if the notes come out a mess whenever there's a third person in the room, that's going to create more cleanup work, not less. And for us that scenario comes up probably a third of the time.
Has anyone dealt with this specifically? Would love to hear from people who've actually tested this in a real clinical setting, not just a quiet solo visit.