CIMI MTF Minutes 20130214

Jump to: navigation, search

CIMI Modelling Taskforce - Meeting Minutes

Thursday 14th February 2013 @ 20:00-22:00 UTC


Linda Bird

Gerard Freriks

Sarah Ryan

Harold Solbrig

Mike Lincoln

Cecil Lynch

Joey Coyle

Jay Lyle

Mark Shafarman

Patrick Langford

Stephen Chu

Eithne Keelaghan


  • Weekly News & Updates
  • Review of Demographics Comparative Analysis spreadsheet & models
    • Party, Actor, Role
    • Person, Organisation

Weekly News & Updates

CIMI 'Terminology & Tooling meetings' @ Tuesday 20:00 - 22:00 TC

Future meeting plans:

  1. Thursday 21st February: Continuation of comparative analysis work
  2. Thursday 28th February: CIMI reference model - Proposed updates and related topics
  3. Thursday 7th March: Model mapping and transformations
    • Results4Care transformation method ?
    • NEHTA template-to-CDA method ?
    • Gerard to suggest? (e.g. Valencia & Seville & Vienna) ?
  4. Thursday 14th March: Model instances; and Immunization model planning

Detailed Minutes

Linda: Happy Valentine's Day! The Draft Agenda...

[Linda goes over Draft Agenda above]

Linda: Next week - continue with Comparative analysis. On the 28th of February, we'll look at the CIMI reference model along with proposed updates and related topics. On the 7th of March - Model mapping and transformations. I need to invite people. If you would like to present, please tell me. After that [on the 14th] - Model instances and Immunizations. Comments?

[no comments]

Linda: Today - continue with Comparative analysis spreadsheet. Start with demographics. Previously we went over address, contact, name...

[Linda brings up EA on the screen]

Linda: Want to show the current CIMI Demographics Model.

[Linda shows the CIMI Participation Model with Locatable]

Linda: A party can be an actor.

We have roles and actors - generalized into party.

So what properties do roles and actors both have?

And what attributes does actor have.

And what attributes does role have?

So - start with parties. LRA... HL7... can be a role or an entity... They all have an identifier and identifier has a value. Done as a cluster because additional details in party. And there are 0-to-mny identifiers.

Stephen: Sorry I am late.

Linda: Stephen - we are planning on meeting on model transformations. Can you present at the meeting?

Stephen: Not yet. I will explain by email.

Linda: OK - Back to party model.

All have name... done as cluster... specified into more detail.

Mark: One correction on RIM - The identifier is set 0-to-many.

Linda: Yes - I have as a Set of identifiers, which means it can have many... but I can change this to a simple cardinality if you'd prefer.

Mark: Should be noted...

Linda: Yes - the cardinality between set and straight identifier is hard to compare... you're right, Mark. The same situation with address... repeatable, only 1 collection...

So each has address... role address... MOHH has...

For contact details, LRA has...

RIM has... which is tel. MOHH has a Tel...

For party type, LRA has a... RIM - could you explain class code and code?

Mark: The 'class code' tells you generic type or entity. The 'code' tells you specific type. So, for example, class code - physician. Code - gastroenterologist.

Linda: OK. Thank you. In Singapore...

Mark: And RIM also has participation Type. So, role of Surgeon participation as a licensed healthcare professional.

Linda: Yes - and participation type is what maps to participation function in CIMI model. So fairly well aligned - is good news. Other comments?

Mark: The chart does not refer to the participation-type code.

Linda: I don't understand what you are saying. Participation type - mapped to Participation.function in the CIMI model. Is that OK that it is in participation class and not here?

Mark: No - that is OK... as long as not left out.

Linda: OK - Looking at party portion in Mindmap.

[Linda brings up Mindmap on screen]

Linda: ON right-hand-side (RHS), the identifier with values and ability to extend. I have done the name and cluster... The name gets more detail as specify. With the address... the address cluster. With contact - electronic contact, which we discussed. Party-type... The only other is the Link... can be archetyped and have link. And the parties can have relationships with other. Maybe need to archetype so have... such as datetime range...

Linda (cont'd): Any comments on Mindmap? No - OK. Can always come back to this.

Linda (cont'd): Next - Look at Actor to see if have additional... that parties do not have. LRA entity... HL7 RM - Entity... ISO 13o... - identified entity. OpenEHR had party-relationship. These were the main ones... They inherit the attributes that we named in party. There must be... because there was one that introduced a... It inherits an address from party, and contact-details when move from party to actor... which restricts the... And also v3 RIM... We need include a descriptor for all actors. Mark - you can advise if need to add these. First was quantity.

Mark: Quantity is for entities when it is important to know the quantities, such as drub dose. Can also be number of persons... but mostly non-person.

Linda: So we have actors that are quantities and we have actors that are groups.

Mark: Two parts to devices. One is ordering device... tracking...

Comment: We can look at specializations. I would say in operate room, number of devices is important.

Linda: Next is date-time range. We were grouping birth-data and death-data. We could pull out and map to date-time range. But for the moment, I have kept these as birth-date and death-date. I'll show you... And death-data has a date-time... I wanted to include, would have to remove date-time from death and birth-data. Would mean that these would be separated from the other birth and death data, which I think would be messy. If people feel strongly that we need to do this, then we can, but I would prefer to leave it as is. Any thoughts? Is it useful to keep as a generalized date-time range for all actors and then specialize when organization?

Mark: It does appear with starting or ending of organization. In RIM there is an existenceTime on all Entities ... Both organization and Person, or types of these.

Linda:So is existenceTime used for Person?

Mark: Probably not used for a person since is a living subject with birth and death time.

Linda: OK - are you comfortable with leaving these as birth datetime and death datetime?

Mark: Yes - how we did in RIM...

Linda: So - what about RiskCode and handlingCode. Are we okay to exclude these?

Mark: In RIM, it is at Entity level, keep... And depending on your politics and... could be risky.

Linda: What about handling?

Mark: At Entity level. Mostly used in Materials.

Linda: So all OK with leaving these out of the Actor, but adding into the specializations of Actor when required?

Jay: In FHIM, there was a discussion about the boundary between demographics and Clinical information - How much is clinical, and how much is for administrative purposes.

Stephen: [Broken up] Not everyone has same type of Sex or Gender recorded.

Linda: Good point, and something else raised.

Stephen: Nicholas at Scottsdale was saying that Sex should be a clinical observation, rather than administrative.

Linda: OK - will revisit. Give... if go to Mindmap... CIMI actor...

Linda (cont'd): Blue highlights = constraints on party. Highlighted - inheritance. Actor Type is specialization of party-type, so will add a restriction on that. I think reason language was added - was originally in openEHR Reference Model. And I think recommendation was to remove... on an entity and move into archetype.

Mark; I am not sure about that. If want to support multiple languages, names have... lots of things use different languages.

Linda: Yes - think about archetype. When specialized down to person, there is a lot of additional details to record about each language, including whether translation required, which is preferred etc.

Mark: I am not arguing that, but other types of devices...

Linda: I need to look into that - will come back to that.

Linda: So other comments on actor Mindmap?

Stephen: One comment on the spreadsheet... there is a name value and a name identifier(?). When you see it is... from description. I see the name-identifier is the id and name-value is the full name. Might want to say name-structure and name-string.

Linda: Good point. The reason I have name-value is that I was looking for a consistent way of people knowing where the whole value of cluster was... and in some cases the whole value. But perhaps we can call as you say. Name structured and name unstructured.. So I agree, Stephen. We should revisit... whether we should make that value different.

Linda (cont'd): So, address is another - full address and address. Contact name... I was hoping... clustered... look at full-value... electronic...

Stephen: I understand, but when look at... you have the name of test and value of test and you go into... But you have name and value here. Naming convention.

Linda: OK - where other place?

Stephen: In other modeling practices, like in HL7 [could not hear the rest]...

Linda: Some are looking at actor and Mindmaps. And Stephen talking about Name value. Please send me an email.

Stephen: OK - I'll send an email.

Linda: Any other models I should look at for role besides LRA and RIM?

[no answer]

Linda: OK. LRA had extract id for... and in reference model, there is a UID. Next is relationship and... There is a party-relationship, and inside relationship there is a source and a target and a type in LRA. And in v3 RIM - a source, target, type. Next is time - validity.

Mark: The role-link has a unique identifier. It is a DSET of II -- id:DSET<ii>[1]

Linda: OK. So performer - is the actor who plays the role. In v3RIM, is the relationship. In LRA... called Entity in LRA - the performer of the role. And [to Mark] - do you have the name?

Mark: Depends on if defined link between 2 roles... Central office controlling... office...

Linda: Not the relationship. Is role to entity.

Mark: We call scope and player. Playing the role... U.S.... an organization.

Linda: Where does scope fit in?

Mark: Not sure of definition of scope in this model.

Linda: OK - I'll open up the RIM.

[RIM on screen - RIM_NormativeContent_v4.pdf]

Stephen: Mark - in RIM - player is person(?) and scoper is organization. Correct?

Mark: Yes. And role class_code... and role code would tell you details.

Stephen: Example of class_code?

Mark: The performer is the participation type. Role [class?]-code might be physician, and code - enterologist.

Stephen: So class_code - treating physician?

Mark: No - treating physician is participation-type.

Stephen: So - class is more general and class_code more specific?

Mark: And participation - whatever formal credential - could be participation as... attending or patient or reading report. Does that give you a feel for that?

Stephen: Just doing...?

Linda: Mark - Role and Participation... Scoper...?

Mark: Yes - a role is something, like a person credentialer... So I can be the entity... in role of physician, or... Participation is distinct from role. Me as patient... You as physician... scope... hospital... physician#... As a physician... you could participate as author... or signer of report. The participation - not as changeable...

Linda: And scoper - organization?

Mark: Yes, or could be a family. When thinking of people... clinical acts... defining the identifier of the role.

Linda: OK - scope is the identifier of the role?

Mark: Yes. Singapore probably has National Identifier. We have thousands - not national yet.

Linda: We have both. Hospitals have their own. OK - so that is something we need to consider in the model. So we have the performer - part of the reference model already...

Mark: The performer would be the participation type. Patient... my doctor would be performer participation-type as discharge-physician or attending.

Linda: OK - I need to re-open window. So - scoper - need to introduce in model. We have the performer as a Party.. But makes sense to add scoper.

Mark: Yes - I think so.

Linda: So - we decided to add scoper... relationship to actor. And performer is now actor. And... (?) - Mark?

Mark: Yes.

Linda: OK. Time-validity... HL7 has effective-time... We have party-type... in LRA, is type, and RIM class code and code?

Mark: Yes.

Linda: Confidentiality level... LRA - only for people. RIM - for whole role... OK?

Mark: If using roles for non-people thing... for therapies, it may be important. A diagnosis might be confidential.

Linda: Not used for diagnosis, but organization.

Mark: So in States, if psychiatrist is seeing you, might be confidential.

Linda: On role?

Mark: Could be on role.

Linda: 0-to-many?

Mark: Yes.

Linda: All OK? OK.

Linda (cont'd): Negation indicator. I know this is widely debated in many circles, and I have seen emails about this, and a paper by Dipak and July James. So CIMI needs to resolve. But for the moment, I have removed from model. In Dipak's paper, he thought it should go into terminology for roles. Comments?

Stephen: The problem raised by terminology was there was no consistent way to do this. This is a problem. Terminology may not be adequate. Done on case-by-case basis - so the question is whether to use terminology or not ... Given this uncertainty, it is up to the individual, but I don't think this is desirable.

Linda: Your recommendation?

Stephen: Constructural constraint, but no consensus.

Gerard: 13606 has an attribute in model to signal that something is going on to signal need for human intervention. Second - negation - handled at the structural level, CIMI may not choose that level. So far, nothing is fixed in 13606. This will be decided this year or later. But I disagree with Dipak.

Mark: There is negation indicator on role class and participation class. In participation - the negation is an indication that the party should not or did not participate in role. So we have cancelled a role for a person - not allowing. A different role. Question is whether CIMI needs to include this at each level.

Linda: We have not had a Use-case for it yet. Include only if Use-case. If we intro here, we need to be clear on...

Mark: Yes - because here it is not negating a terminology value.

Linda: Yes - once get to not participate in something that also has a negative, complications...

Mark: Yes - that is why the RIM has it on each of the classes. So if you have a collection of acts in a document, and a party participates in first, third and not fourth...

Linda: So negation of role - do you have an example of that?

Mark: Specify that role is a competency... contributing to entity... Example: This person is not our employee. This mouthwash does not have alcohol as an ingredient. Could say this person is not a doctor at this hospital.

Mark: Is about... negating a role... or person hasn't participated in particular act.

Linda: I suspect we will eventually need to add, but tempted not to add until Use-case. Are all OK with this?

Stephen: Yes - fine with that.

Linda: I suspect we will need to re-visit because will have Use-case. OK - next is status of role. Status role in RIM. And a number of different attributes from RIM. I have tentatively decided to exclude until we have a use-case for this. So what about Quantity - is this needed here, or in one of the specializations of party. Any thoughts?

Mark: Probably you will need quantity when get into ordering and probably licensing... when get into... but probably not need now.

Linda: And position and priority?

Mark: Same thing.

Linda: So - leave out. Next - scoper and confidentiality. Mindmap - Confidentiality-level is 0-to-many. We talked about: The identification inherited from Party, Name Cluster inherited from Party, Address inherited from party, Contact, role-type, date-time range, Confidentiality-level...

Jay: Why confidentiality level?

Linda: because of what RIM does.

Mark: In States - certain treatments added for a person in certain roles.

Linda: If can have different types of confidentiality levels...

Jay: Confidentiality... access to information or level of confidentiality?

Mark: To identify... how sensitive piece of information is or...

Linda: Is that role?

Mark: ...

Jay: So is both...?

Mark: In medication, could be certain drugs, or in clinical trial... or in an act... psychiatrist.

Linda: So confidentiality?

Jay: For viewing or accessing?

Mark: Is info itself, or how info can be made available.

Linda: An indication of the level of confidentiality required for accessing information associated with this role?

Mark: Either the role information itself, or...

Stephen: For any particular piece of information, you should only be associating one level of confidentiality...

Jay: Could we get one concrete example... people... fact?

Linda: When role is scoped by a particular organization. So at hospital, anything associated with role, certain confidentiality?

Stephen: That is only 1 confidentiality ... one value... one piece of information?

Linda: And would need more fine-grained.

Mark: Exactly.

Jay: Still only hearing one. Why would I have two if a single role?

Stephen: I have same question. Maybe a disclosure rule for payer or different rules for other providers...?

Mark: Could be different confidentiality-levels for different... on different roles.

Jay: Example?

Mark: Normal restricted substance-abuse related... (?)

Jay: OK.

Linda: Other comments on role model? OK.

Linda (cont'd): We've done actor and role - two top-level classes in model. We have twenty minutes left - won't get through all of person. But person-name is interesting.

Linda (cont'd): So person as actor... FHIR has person... LRA has... HL7 has Living Subject and person... ISO13606 has person... 22220 has subject of care... DCM has person... FHIM has person/biological entry... NEHTA has participation-person... The first category is the relationship. Actor or party can have... So person can have relationship to healthcare provider. So healthcare provider relationship is openEHR... to... and in ... and in DCM...

Linda (cont'd): We have the source which is the person... the target... the provider... relationship-type... Joey?

Joey: Yes.

Linda: I need to follow-up... need to go to and open-up?

Joey: Yes.

[Linda opens]

Mark: In RIM, it is usually a player and a scoper of the patient-role. So-and-so is the patient of so-and-so. It is not really role-link. That is one role has formal or legal authority over other roles.

Linda: OK - I will need to revisit before next week. Question - whether healthcare-provider relationship should go on healthcare consumer role, or stay here on Person. We will get to this when we look at healthcare consumer role ... So let's look at name.

Linda: Inside person-name, a whole bunch of attributes. In some of the models... 22220. Gerard - was interesting. Some models specialized middle name... Professional title... Nobility title was interesting, Joey. History of this at Intermountain?

Joey: I don't know.

Gerard: Dutch... separate from family name. It doesn't have to be related to nobility. Sometimes... from a place...

Linda: English translation?

Mark: You do see in Dutch and German names, like Ludwig von Beethoven... the von...

Gerard: You see in French, too. Relationship with nobility and places...

Linda: That is why I put nobility, but family name as well.

Gerard: Is part of family name - related to family name. My name could be Gerard Freriks from Amsterdam. Nobility is separate - that is the prefix.

Linda: I'll just have to call VoorvogelsName

Linda (cont'd): Following the 22220 name structure, then have a family-name group. So if we follow that...

Jay: Wikipedia calls it a nobilitary particle ...

Gerard: I think family-name-prefix...

Linda: I like that. Much easier to pronounce. Are all OK with that?

Gerard: I call it the prefix-name group. Or prefix-family-name group.

Linda: These sequence numbers seem to be a sequence within that group, so if we sequence it... So I think we need to group with family name.

Gerard: The way I interpreted... name-title-group and... and then suffix-name-group.

Linda: So this goes with family-name-group?

Gerard: Yes, and can have a family-group... Can have many family names. The last time I looked at it...

[Linda looks at ISO TS 22220 diagram]

Linda: ... family-name group with sequence of 2... Good - that is also my interpretation.

Linda (cont'd): So... with address... can take general person name and specialize into details...

Linda (cont'd): So - some have given name. I have separated in spreadsheet, but I could say middle name is type of name... IMH. We have family names... We have family-name prefix. We have name:title - specialized by the type of title in some models. I've mapped to IMH... I'll check.

[Looks at CEM Browser]

Linda: Is that the case, Joey?

Joey: I don't know about person models, I did mostly tools...

Linda: OK. We should probably be... Mapping up to... In IMH - there is a prefix and title. In other models, the prefix was the title. So - this is Mr. Ms. Dr. Van - interesting. And title - more like academic title... Is that something you could get more information on?

Joey: What is question? What is difference between the two?

Linda: Yes - Prefix name. Mr., Mrs., Ms... Whereas your IMH title name was a special title

Joey: I think... one's how would write name out... and the other...

Linda: But, professor...

Joey: I'll ask Stan.

Linda: I've mapped prefix to title and title to Professional Title... for the moment. So - we have the name suffix. And... Name:use. Datetime Range of Name:Use. And the Usage Conditions. Status of name, and whether it is preferred or not. Abstract name model... and details. We'll pick up next week. Comments?

Stephen: I will send you 4846 - some info about title and prefix. Australia - combination of Mr. and Mrs. So if person has professional title... Then known as... Then there is family prefix... The Australian standards...

Linda: Interesting.

Stephen: My take on this... I wonder... They have suffix and prefix... that is how this organized... My understanding... Correct if I am wrong...

Linda: Interesting - that is why I tried to capture as professional title. Can still be... IMH.

Stephen: Only constraint on 4846 - please do not circulate. Now allow to circulate. Also - if present on transformation as well... Follow-up.

[end of meeting]