Notes #1556
Replies: 3 comments
-
A starting point can be: https://github.com/synthetichealth/chatty-notes This is a set of python scripts that take the Synthea FHIR Bundles, pull some of the information and send it to the ChatGPT API to generate more narrative clinical notes. Some caveats:
|
Beta Was this translation helpful? Give feedback.
-
That's awesome!
Where's the prompt? I don't see it...
This could be very useful.
…On Mon, Feb 3, 2025 at 3:39 PM Andy Gregorowicz ***@***.***> wrote:
A starting point can be: https://github.com/synthetichealth/chatty-notes
This is a set of python scripts that take the Synthea FHIR Bundles, pull
some of the information and send it to the ChatGPT API to generate more
narrative clinical notes.
Some caveats:
- This was my first generative AI work. It uses a "one shot" prompt
and is not the way I would do things now, but the code could still be a
useful place to start and/or may be good enough for your purposes as-is.
- It was tested on the ChatGPT 3.5 API. I don't think API access
existed to 4.0 at the time. API access code may be totally out of date.
- The code has special cases for osteoarthritis as that was the focus
of the work I was doing.
—
Reply to this email directly, view it on GitHub
<#1556 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/BMFJMRD74JAZIRRSAASLIQT2N7HYLAVCNFSM6AAAAABWM6TOPOVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTEMBUG4ZTQMA>
.
You are receiving this because you authored the thread.Message ID:
***@***.***
com>
--
Will Bartlett, MS, MBA
646 335-2836 | Calendly <https://calendly.com/will-h-bartlett>
|
Beta Was this translation helpful? Give feedback.
-
Prompts are in the templates directory: https://github.com/synthetichealth/chatty-notes/tree/main/templates They are Jinja templates. If you look at chatty.py lines 179 - 193, you'll see how it looks at the FHIR data to select a particular template. |
Beta Was this translation helpful? Give feedback.
-
I need realistic patient notes for my project. While some notes exist in the data (I'm using Coherent), I have found that they are mostly rudimentary restructurings of information contained in the structured data. Is there any way to use this dataset or software to access / generate richer, realistic notes, or is that out of scope?
Thanks.
Beta Was this translation helpful? Give feedback.
All reactions