Hi everyone / anyone 
Very happy to join this forum having been interested and drawn towards OpenEzhR for a while, thanks to @Kanthan_Theivendran, Ian McNicoll, Alastair Allen and for helping me to better understand it and also an excellent 13 minute explainer video from @Sidharth_Ramesh actually which very much helped it all click into place for me.
We are looking at building a health app for GPs and lower tiered cities in India, helping improve transparency, efficiency and access to better quality healthcare and am exploring doing so on OpenEHR.
Iād like the app to utilise speech to text / voice, but I have searched āvoiceā and āspeechā here in this forum and thereās no mention of this that I can find.
Has anyone built UI/UX within the OpenEHR tooling/framework that handles this? Iām not sure if this is viewed as an āupstreamā factor and OpenEHR will handle any converted voice once itās in text, but the UI and API considerations in terms of making this work well/fast/effectively from a UX standpoint is where I am comimg from, I guess. Can this be done using existing UI extensions / SDKs or would we need to develop this?
More generally, anyone who has or is looking at this, would love to hear from you, please
Thanks!
Olly
2 Likes
@Olly_Cogan I saw in EHRCon24 @Sidharth_Ramesh showing a demo of this functionality. I also know some others are working on it but canāt disclose as it may be confidential so Iāll let people jump in here to answer. Pretty cool functionality which would significantly reduce burden on providers.
I had my own little attempt to hook up an openEHR CDR to voice, albeit patient input not clinical
https://nhshackday.com/projects/23-cardiff/the_ballad_of_yann_and_alexa
Worth confirming though that openEHR itself is only about the data layer and definitely ādownstreamā of this kind of application, which is why you will not anything in official openEHR documentation.
There is nothing special about the openEHR API that would make it unsuitable to work with text/voice analysis which presumably , initially produces narrative.
Then we are into the challenge of generating structured input from narrative.
4 Likes
Hey @Olly_Cogan I can share a demo of the work weāve been doing.
Weāve got a generic voice ā forms converter that takes any openEHR template as input. Thanks @Koray_Atalag for the mention.
Weāre currently experimenting with integrating external terminology referenced in openEHR templates. Eg: terminology://fhir.hl7.org/ValueSet/$expand?url=http://hl7.org/fhir/ValueSet/allergyintolerance-code in a DV_CODED_TEXT field as a refset should make the system only output terms from the terminology system.
I canāt share a video demo in public yet, but if you email (sidharth [at] medblocks [dot] com / dm me, I can show what weāve been doing.
2 Likes
Hi all! We are setting a research study with @jpieraj to validate the population of openEHR templates from a clinical conversation in a real clinic using LLM and probably agentic RAG. It is early days so we will keep you updated on how it goes as we have some challenges such as Catalan/Spanish language use during the consultation, etcā¦ Will contact you @Sidharth_Ramesh to see how we can collaborate on this!
4 Likes
A very important and interesting initiative - well done! I believe that Apperta is coordinating a similar kind of exercise in context of their efforts towards OpenEyes/openEHR harmonisation. I understand that we may have results demonstrated soon - perhaps at the January OpenEyes development group meeting.
4 Likes
@Olly_Cogan Thereās a public demo of the speech to openEHR compositions running as a SMART app now available online: https://youtu.be/ShCzrHlzmHo?si=MXQVhoTo1yfuIJ6U&t=969
Go to around 16 mins and 10s to see it. @sebastian.iancu, @xabiermichelena and @Koray_Atalag have already seen 
4 Likes