As a CKA, I’m always pleased to see the archetypes that we develop as a community welcomed and used within systems.
However, there seems to be a significant mismatch between the relatively small number of high value, core reusable archetypes (my bias ) that have gone through the publication process. There must be many more vendor-developed archetypes that have enabled sophisticated clinical functionality to be built into their systems.
Is there a reason why they are not being proposed to the international CKM, so that they can be shared more broadly?
What can we do to improve the situation?
Hi Heather, at Nedap we build many archetypes, most of them are clinical scores. List here (in Dutch) NB only the ones listed as ‘in productie’ are actually ready for production. Airtable - Grid view
Reasons we haven’t proposed them to CKM yet are multiple:
not sure there’s any interest
Dutch language only
I thought CKM was only for core archetypes
It’s build using company resources, we’re open to discuss wether sharing can be in our interest
Archetypes/templates I shared before (around care plan) got limited attention.
I think there are multiple locations where vendors share archetypes (mostly github). I’d actually like it to be easier to discover and search those.
CKM is only as good as the archetypes that are volunteered as candidates. Somebody else has usually those created as a result of an implementation or project. The intent is to take the benefit of experience of those who have gone before by reusing existing, especially published, archetypes. But the library of archetypes only grows if new archetypes are shared back. There is no obligation to do so, but… if vendors don’t reciprocate how can the library grow?
The potential for interoperability between vendors is directly proportional to the number of archetypes they share.
Can you explain further?
Why not CKM? In public everyone talks about sharing high quality archetypes, yet in practice…
Yeah you’re right. We agree it’s important to share with the world. It’s why we opensourced Archie. And other non openEHR stuff. I didn’t know there was an interest in our score archetypes. Could you make sense of the airtable list I shared? Please tell mee which archetypes are of interest to the ckm.
The potential for interoperability between vendors is directly proportional to the number of archetypes they share.
Off course I know and support this. But in practice we don’t do any interoperability via openEHR as all the attention in our country goes to FHIR.
About sharing archetypes: we spend serious resources on creating these archetypes. We do it to better our product. Sharing these archetypes means our competitors benefit from our work which can be against our interest. There’s also much to say for sharing: if it’s good for openEHR it’s good for Nedap, it’s good pr, it makes Nedap employees feel good about our company etc. It may encourage others to share as well.
There’s also ways to offset the risk: building trust with our competitors, agreements with competitors, licensing restrictions, sharing only a few archetypes at first etc. Interested in your thoughts:)
Maybe it’s something to bring up at the next industry partner meeting as well @Tomazg ?
We share the clinical models. The github repo is deprecated. We found it not practical to have one big repo with all. Currently we add archetypes and templates to the product repos. They are not public.
We work close with #openehr Norway. They represent our customers. We strive to have reviews with them for all new products. As part of our product development we develop and suggest new archetypes. If #openehr Norway accept them they will be added to ckm. Then there will be national and/or international reviews on them.
I gave Boston Bowel Preparation scale as such an example. We saw the need working with clinicians. Read the published knowledge about Bowel preparation and developed the archetype. This was reviewed nationally and internationally 3 months after the initial prototype. Fantastic!!
There are currently no formal reviews of Templates. This is something we (#openehr community) should look into. It’s in the template the “magic” happens. We constrains, expands (slot filling) and add terminologies.
Well said, @joostholslag@bna . There is ongoing working of lifting/importing archetypes from our open source common modelling git repo to the openEHR CKM, facilitated by the Swedish editorial process in my understanding.
Still there is no support for GDL2 guidelines in the official CKM.
I would think the starting point there would be to get as many of the GDL archetypes into CKM as possibl. Many are scores / scales. Since many of the Nedap ones are also scores, it would seem worthwhile considering performing a cross-check of some kind e.g. at least the names, to see what the overlap is, and then identify some good ones / highly reusable ones, and translate to EN and propose to CKM.
These archetypes have cost Nedap, Cambio and anyone else who does them hundreds or thousands of hours. So why share them? Well, … good PR etc, as @joostholslag mentioned above. But there is also some enlightened self-interest. If Nedap shares their DEMMI (which I just learned was Morton Mobility Test!) and it gets taken up, then all data in openEHR systems for that test start looking like the Nedap model. Might be good if Nedap has an app that does interesting stuff with this particular data… now it can be used in more places. Same for the 500+ score/scales from Cambio - if at least some of those get into CKM as the accepted version of that scale, then Cambio apps or CDS that know about those data sets will work in a lot of places. And so on for DIPS etc etc.
The big picture: no matter how amazing any one company’s archetype repo is, it’s probably only 5% of medicine. CKM might be ?30% coverage - and it’s already the largest clinical model repo in the world by data point count. But if all those 5% libraries get added together into CKM, then we’re getting serious coverage, and each company gets for free many times its own repository worth of archetypes.
Another thing: there is nothing wrong with charging customers (typically healthcare providers or authorities) for the development work on those archetypes. There is also nothing wrong with telling them you will open source them, and everyone does that - so that they will not get charged anything for the greater number of archetypes already built.
I have no knowledge of what is happening by the Swedish editorial process but over the years, with Cambio’s permission, been slowly working through the GIT repository.
When I last looked, early 2021, there were 404 archetypes. Of these:
25 (6%) archetypes are suitable for inclusion in CKM
** Of these, 14 (all OBSERVATIONS representing scores, scales or indexes) have been worked up according to Editorial guidelines and imported to CKM.
** A further 11 of the 25 have been identified as candidates but need further investigation or Editorial work before they can be uploaded.
83 (21%) have been rejected for various reasons - the main reason being use-case too specific and not designed for reuse or concept scope eg specialisations per test result or TNM disease or medication
140 (34%) are not eligible to be imported as they are EVALUATIONs built to carry an assessment outcome etc that should be included with the matching score/scale OBSERVATION archetype. If an OBSERVATION has been imported into CKM then the matching EVAL is found and the contents added to the OBS.
29 (7%) are sourced from CKM
126 (31%) yet to be analysed, but anticipating we will find more suitable for upload
This repository reflects the messy real world of modelling and teaching students and we need to be mindful that the original purpose (as I understand it) was to support GDL experimentation. These archetypes were never modelled with an aim to upload CKM, and that is perfectly OK.
While not all 404 archetypes are suitable for CKM, we have been able to upload some gems that are now available for reuse.
I’m grateful as a CKA and the whole community benefits, in the same way that existing archetypes supported some of the work done in this repository.
BTW all of these archetypes uploaded to CKM carry both the original author’s name and shared copyright between Cambio and openEHR.
From my POV it has been a great example of collaboration and sharing resources.
just a quick input from my side: vitagroup currently works closely together with partners in HiGHmed for various projects (NUM-CODEX, Nephrodigital, CaEHR). Modelling is done by documentation specialists from the HiGHmed university hospitals and made available through the HiGHmed CKM.
You might want to take a quick look at the Nephron-Digital project which contains some models for dialysis and Banff-Classifications: Clinical Knowledge Manager
I can ask my colleague @nina.schewe to make these archetypes candidates for the international CKM if you consider these as suitable.
On the wiki-page “CDS-arketyper” some Swedish clinicans (at least @ark and @Anders_Thurin - maybe others too) involved in the national Swedish openEHR collaboration have made a prioritized list of archetypes from the Cambio repository that they believe most useful.
Many of them either come from the CKM or have already been included as @heather.leslie describes above. The sentence “Samma som i CKM” in the list means “Same as in the CKM”.
I believe the intention is to start informally reviewing and polishing at least some of the other archetypes and then work through the normal international CKM and Discourse channels to suggest/discuss future publishing in the CKM.
@erik.sundvall’s description of the work is entirely correct. I have been struggling to get some time to restart the work for some weeks now, but unfortunately didn’t find the time this week either. I hope to be able to restart soon.
@heather.leslie; If you have found out any specific archetypes that would be useful to polish and then upload to the CKM you are more than welcome to point them out to me.
I started this while Program Lead, but it wasn’t completed. Here is my working file, as is. There may be some minor errors of status but it might save you some work, so happy to share.
The analysis is based on the archetypes that were present in the repository on 30 September 2020, so there may be more now.
The stats in my previous message apply to this spreadsheet:
Orange are candidates - need Editorial work or further investigation for content clarification etc
Red - rejected, as described above
Crossed out - already exist in CKM
Uncolored - not yet analysed.
The early modelling patterns seem to be building OBSERVATION/EVALUATION pairs - the OBS carrying the ‘facts/evidence’ and the matching EVALUATIONs built to carry an assessment or interpretation. These need to be merged into a single OBSERVATION, if the interpretation is considered to be a formal part of the score or scale.
There may be some scores or scales that have been incorrectly modelled as EVALUATIONs, and need to be transformed to OBSERVATIONs so each ADL file has to be manually uploaded into the archetype designer and viewed, edited etc. It has been fruitful, but surprisingly time-consuming.
Regarding the demographics archetypes associated with CHADSVASC - might be worth taking a look at the demographic archetypes currently under review to see if they are useful for this purpose.
Unfortunately, now that there is no dedicated and funded openEHR modelling resource to facilitate the review process, relying on volunteer time is likely to mean that publication processes in CKM are likely to slow down. Skilled Editor capacity is limited, and CKA availability is on a volunteer basis.
) that have gone through the publication process. There must be many more vendor-developed archetypes that have enabled sophisticated clinical functionality to be built into their systems.
Why not CKM? In public everyone talks about sharing high quality archetypes, yet in practice…