About Public Involvement in Healthcare / Sur la participation du public dans le soins de santé
    ai scribes ai evaluation
    Loading More Posts
    • Oldest to Newest
    • Newest to Oldest
    • Most Votes
    Reply
    • Reply as topic
    Log in to reply

    Evaluating Ambient AI Scribes
    2
    3
    17

    This topic has been deleted. Only users with topic management privileges can see it.
    • C
      Chris Johnston last edited by Chris Johnston

      Saw this article and thought it might be of interest:
      Evaluation of an Ambient Artificial Intelligence Documentation Platform for Clinicians

      Amongst the results, one that caught my eye was:
      Mean (SD) time in notes per appointment significantly decreased from 6.2 (4.0) to 5.3 (3.5) minutes (P < .001), with a bigger decrease for female vs male clinicians (8.1 [3.9] to 6.7 [3.6] minutes vs 4.7 [3.5] to 4.2 [3.1] minutes; P = .001).

      So that after the implementation of AI, despite the bigger decrease, female clinicians were still spending on average significantly more time on each patient note than male clinicians had been before the implementation.

      The study has a number of limitations that may mean it's not generalizable but still - hmm, food for thought.

      1 Reply Last reply Reply Quote Edit 0
      • K
        Kim Locke @Chris Johnston last edited by

        @Chris-Johnston Is there anything mentioned about transcription in the article? I see two specialists that use this form of taking patient notes, and they're paying people to write down/type the notes from their audio.

        This makes me wonder if this could be easily automated with AI, and then have a human look over it quickly for any transcription errors?

        1 Reply Last reply Reply Quote Edit 0
        • C
          Chris Johnston @Kim Locke last edited by Chris Johnston

          @Kim-Locke Hi Kim, it doesn’t specifically mention transcription but that is exactly what ambient AI is designed to do - it listens and generates notes for the patient record. So it’s not generally producing a verbatim transcription, although it’s capable of doing that, more often it’s creating a summary of key points discussed and consensus on next steps in the care plan.

          For anyone recording conversations directly, it’s entirely possible to simply choose an AI to upload the audio files and have it transcribed. (Searching “AI to transcribe prerecorded audio” produces quite a few options). And of course for virtual conversations over platforms like Zoom, there are both built in and third party AI assistants who will transcribe or summarize the conversation directly.

          However - and this a very big caution - while it’s possible, doesn’t mean it’s recommended. Uploading confidential patient consultations to an AI tool would of course be a breach of privacy and ethics.

          The physician’s practice would need to enter into a contract with an appropriate vendor to safeguard patient privacy, in compliance with provincial and federal legislation, by guaranteeing that recordings are handled confidentially, not transferred to or stored on servers outside Canada, destroyed once transcriptions have been approved etc, etc.

          But there are plenty of initiatives out there to help physicians do exactly that including a national initiative that will be explored in an upcoming webinar, see here for more info:

          18 June Webinar: AI Scribes & New National Initiative

          Hope that helps 🙂

          1 Reply Last reply Reply Quote Edit 0
          • 1 / 1
          • First post
            Last post