Archetype authoring attribution

Hello,

Back again with the licensing topic of archetypes, with a real use case.

We have been asked to help in creating a set of 13606 archetypes for breast and prostate cancer. Although they will probably incorporate some new requirements, the main source will be some of the openEHR archetypes available at the CKM.
Assuming that the have adopted a CC-BY(-SA) license (I cannot recall which is the state of that discussion), the doubts are the following:

  • Converting the archetype to a new reference model is considered as a derivation? Or the openEHR archetype is considered just as a reference material as could be any textbook or paper?
  • The author of the new archetype has to be the one of the openEHR archetype (Ian McNicoll btw) or the person who in fact creates the new RM-based archetype?

The underlying question here that should be clarified is to define which is the extension of the concept “derived work”.

David

David,

openEHR:
Creative Commons license CC-BY-SA 2.0, applying to all openEHR.org achetypes hosted at the openEHR Clinical Knowledge Manager (CKM).
http://creativecommons.org/licenses/by-sa/2.0/

My simplistic understanding.

A derived work has to be derived.
So when you use the information and transpose it as constraints on an other RM,
then I consider this new archetype as derived from that new RM it is transposed to.
So when this approach is followed then the Attribution is to the group that provided the clinical content.
But there is no attribution to the openEHR RM specification.

When you translate the text in the openEHR archetype to Dutch it is derived but still derived from the original openEHR RM.
In this case attribution must be stated to openEHR RM and the clinical group.

Is this an answer?

Gerard Freriks
+31 620347088
gfrer@luna.nl

Hi David

I am tied up at a conference all day but will respond in more detail later. This is a very important question and I see no barrier to you adapting these archetypes to 13606 but I would much prefer to collaborate in extending the openEHR archetypes. We have not made a final decision on the licensing of archetypes but the alternatives being considered are even less restrictive to reuse, derivation.
Ian

Dr Ian McNicoll
Clinical modelling consultant Ocean Informatics
Mobile +44 (0) 775 209 7859
Skype imcnicoll

Hi David

As the aim for all is interoperability of these things, I would hope that the information would be two way. I would suggest getting the new experts to comment on CKM and then derive a 13606 archetype (this is described in the 13606 standard). I would like that to be a future part of CKM but understand this may seem a little too controlling.

If we start creating clinical content specifications in lots of places it will not really assist medicine a great deal. We estimate that it is costing health care dearly to do this again, again and again. Particularly when providers are interested in quality and sharing information.

That said, I would attribute the work to openEHR, the original authors, contributors and any new expert inputs. The license is to openEHR so I guess it is openEHR that needs attributing if you want to stay with the legal requirement. The SA does mean that you have to share the derived work under a similar license, something that some have been worried about. I am interested in your views on this.

Cheers, Sam

if it is the same archetype, then it is a derived work. Which is fine, that's what CC-BY is for. My understanding of the term is that a machine conversion to another format (which is essentially what you are saying) would be a derived work - legally not different from JPG -> PNG I suspect.

- thomas

Hello,
See in-line.

2012/3/22 Sam Heard <sam.heard@oceaninformatics.com>

Hi David

As the aim for all is interoperability of these things, I would hope that the information would be two way. I would suggest getting the new experts to comment on CKM and then derive a 13606 archetype (this is described in the 13606 standard). I would like that to be a future part of CKM but understand this may seem a little too controlling.

If we start creating clinical content specifications in lots of places it will not really assist medicine a great deal. We estimate that it is costing health care dearly to do this again, again and again. Particularly when providers are interested in quality and sharing information.

Of course, sharing these models is the only way to achieve global agreements I do not know which kind of changed they would require for the archetypes (if any) but I think they will be related to the real data available at the EHR system of the hospital. So they will be probably very localized changes. What clinicians have done right now is to translate most of the archetypes into Spanish, and that can be very easily incorporate into the openEHR CKM archetypes.

That said, I would attribute the work to openEHR, the original authors, contributors and any new expert inputs. The license is to openEHR so I guess it is openEHR that needs attributing if you want to stay with the legal requirement. The SA does mean that you have to share the derived work under a similar license, something that some have been worried about. I am interested in your views on this.

Cheers, Sam

There is no doubt about the attributions and original references that must accompany the new archetypes (by the way, maybe in this sense the archetype metadata could be improved. Diego Boscá has been working on this topic for his PhD). The question as I said before is about the authorship attribution and the meaning of “derived work”. See below.

2012/3/23 Thomas Beale <thomas.beale@oceaninformatics.com>

if it is the same archetype, then it is a derived work. Which is fine, that’s what CC-BY is for. My understanding of the term is that a machine conversion to another format (which is essentially what you are saying) would be a derived work - legally not different from JPG → PNG I suspect.

  • thomas

Probably the problem is not so simple. I will put different options of things that can happen as an example (any new case is welcome):

1- If I take an openEHR archetype and modify/specialize it as a new openEHR archetype it is a derived work.
2- If I take an openEHR archetype and generate an implementation guide document of it, it is a derived work. The change of the format does not affect as you said.
3- If I take an openEHR archetype and generate software, schemas, etc. as Thomas said in a different thread they “are not derivative works, they are original works based on the specification”
4- If I take an openEHR archetype and generate another archetype of a different reference model based on it (could be 13606, HL7 CDA or whatever), is this a derivative work? The fact that the openEHR to 13606 conversion is nearly straightforward is not relevant here. It could be not the case. At the end someone (or some automatic process) will have to decide the correspondence between different reference models. For me this is exactly the same case as point 3. Thus, should not be considered a derivation but a new work which uses the original archetype as a reference, as could have been any textbook or paper.
5- If I take an openEHR archetype and generate an HL7 CDA implementation guide based on it, is this a derivative work? The answer to this depends on the previous one. The fact of representing clinical models in a different format (if we see ADL just as a format for defining models) should not change the essence of the problem as we saw in point 2.

See that I’m just trying to set out the limits of the problem to find a general rule if it is possible.

David

I would be very interested to see proposals for improvement of the meta-data structures. We could get these into ADL/AOM 1.5

- t

Hi David,

Firs of all, leaving aside licensing issues and the eventual RM choice, I would really appreciate working collaboratively on these models with your clinical guys. I would be happy to do this via CKM if you felt that might facilitate progress. As Sam says, lets make sure we keep the clinical models consistent even if the final formailsm differs. I don’t think the usual 13606/opeEHR philosophical differences have much impact in the histopathology space and when I did the original work for the RCPA, I was impressed at the level of international cooperation that exits in the community already. In hindsight, I would have modelled some of the histopath content a bit differently and since none of these archetypes are formally published, we do have an opportunity to learn from that experience + of course include any new requirements that your clinicians identify.

With regard to licensing, firstly I agree with Sam that attribution is to the Foundation, and not to me personally. By all means send me a large cheque if that makes you feel better :wink:

The reason for the current CC-BY-SA licensing is really only to try and prevent restrictive re-commerciaisation of archetypes that were originally developed withan open licence.

There is no within the openEHr community about what we are trying to do. The differences of opinion are about how best to achieve this goal with the minimum of restriction

So, the principles we are trying to work to are that …

  1. You are perfectly entitled to adapt or derive any openEHR archetypes
  2. You are perfectly entitled to use these in a commercial application
  3. Personally I am not bothered if you try to sell them (that may not be a consensus view).
  4. We are definitely bothered if you try to assert copyright and attempt to restrict others from using or adapting the original archetypes or derivatives of those archetypes.

It is (4) that is the key problem here and archetypes live in a tricky limbo between software artefacts and human document which means that legal opinion is not well-based on precedent or evidence and making a decision on the best licence is tricky.

Personally, I take the view that, as a clinician, the work that I did was simply a restatement of universal clinical knowledge and as such must really stay in the open domain. One of the problems in this area is that CAP (College of American Pathologists) have done some similar histopath work, in particular defining SNOMED bindings but have locked the IP away. I think this is against the spirit of
clinical knowledge development, though actually I know how to access the bindings via the public domain :-).

I really do not know whether the transformation from an openEHR archetype to 13606 would be still be considered ‘derivative’ by a court of law but if we get into that world we have all lost the plot, in my opinion.

To sum up. From a personal perspective and my interpretation of CC-BY-SA and the Foundation objectives, feel free to use anything you want. Unless you have a cunning plan to commercialise the archetypes and then lock the rest of us out, I do not see any problem at all, wether or not your work is regarded as derivative

But let’s find a way of collaborating - would you be interested in working via CKM, accepting that you might want to convert the end product to a 13606ENTRY?

This is all part of a very interesting wider public debate about we reward those who create or add value to content/knowledge without the dead hand of patents and resultant legal dispute.

Regards,

Ian