Select which SOURCES are being used to create an output.
Kate from Heidi
Merged in a post:
Have Heidi use the transcript (not the first note created) as the foundation to create new notes from the same patient interaction
J
Joseph Garcia
Have Heidi use the transcript to create secondary "notes" (i.e., notes that are created after a first note is created), instead of using the first note created as the foundation to create the second note.
Explanation:
When creating a second note after a first note has been generated from a transcript (likely to adjust the style or add information excluded from first note), Heidi appears to use the first generated note as the foundation to create new note (instead of the original transcript). It would be good to use the original transcript instead, as secondary notes seems to exclude details from the transcript if the first/original note did not include those details.
Example:
- If the original generated note did not include a robust HPI with details of poor sleep, any post-requests to make the HPI include more details are limited because it seems to be using the first note as the foundation, not the original transcript.
- Two secondary notes using the same template will have different logical structures to the HPI depending on what style the primary/original note is in.
M
Melina Dion
I get the same issue too and it's really frustrating. The more I try to regenerate a note and adjust it to fit what I am looking for the more it changes and the more it starts hallucinating
Canny AI
Merged in a post:
Allow multiple notes to be made from a single transcription
C
Chris
Sometimes there are things that are discussed that I would like to include in a document, but not necessarily in the medical note. For instance, when discussing estimated prices of services, I would like to create a note that lists what service(s)/procedure(s)/etc. that are discussed and the costs of those, however, those do not need to be listed in the medical record as that is not pertinent to the patients record. In my note templates, I often tell Heidi to exclude any mention of costs (which doesn't always work correctly), therefore, with the way documents are supposed to work, pulling only information from the generated note, an estimate document would not be able to work. I'd like to be able to have my medical record note which does not include costs, and the ability to generate a new note that then contains only the information I ask it to use at the exclusion of everything else, such as an estimate note
J
Jamie McConnell
Whilst this is an interesting perspective, I believe that that having the notes act as the "truth" is the correct solution. The transcript is by its nature unstructured and hard for a person to read quickly. Mistakes are much easier to spot in a bullet-point list.
In fact, I very frequently see Heidi revert to its mistaken interpretation of the transcript, when I have already corrected those errors in the note.
I think what we are both unhappy with is the 'black box' nature of the user interface. It would be much better if Heidi presented a (detailed) summary note, which represented its complete understanding of the conversation. In an ideal world, this would appear in realtime.
The user could check this note for errors and omissions, and then all other outputs (short note, long letter) could be based on that.
This would require some tinkering with the current UI, but it would greatly improve user confidence.
M
Melina Dion
Jamie McConnell I think a solution for that would be having Heidi only use the new information in the note that it did not generate itself (so only the edits the physician made) as "the truth" to generate a new note
J
Jamie McConnell
Melina Dion
Yes, that’s a much better way of looking at it. Heidi should regard anything typed in the Context as being the “truth”, and anything it generates that contradicts the Context should be dismissed.