From: owner-mnm@lists.hl7.org on behalf of Lloyd McKenzie
[lmckenzi@ca.ibm.com]
Sent: Wednesday, July 09, 2003 5:53 PM
To: Gunther Schadow
Cc: Russler, Dan; 'QUADE_LINDA@LILLY.COM'; mnm@lists.hl7.org; Bob Dolin;
Sandy Boyer
Subject: Re: RIM PROPOSAL: Add attribute Act.title

Importance: Low

How about this:

<definition>
A word or phrase by which a specific Act may be known among people.

Example: name of a research study (e.g. "Scandinavian Simvastatin Study"), name of a court case (e.g. "Brown v. Board of Education"), name of another kind of work project or operation. For acts representing documents, this is the title of the document.

Discussion: This is not a formal identifier but rather a human-recognizable common name.  However it is similar to the id attribute in that it refers to a specific Act rather than a 'kind' of act.  (For definition mood, the title refers to that specific definition, rather than to a broad category that might be conveyed with Act.code.) </definition>


Lloyd McKenzie, P.Eng.              I/T Architect, IBM Global Services
Internet: LMCKENZI@CA.IBM.COM
PhoneMail: (780)421-5620          Internal Mail: 04/QBG/1004 /EDM


Gunther Schadow <gunther@aurora.regenstrief.org>@lists.hl7.org on 2003-07-09 03:43:31 PM

Please respond to Gunther Schadow <gunther@aurora.regenstrief.org>

Sent by:    owner-mnm@lists.hl7.org


To:    "Russler, Dan" <Dan.Russler@McKesson.com>
cc:    "'QUADE_LINDA@LILLY.COM'" <QUADE_LINDA@LILLY.COM>,
       mnm@lists.hl7.org, Bob Dolin <Robert.H.Dolin@kp.org>, Sandy Boyer
       <slboyer@attglobal.net>
Subject:    Re: RIM PROPOSAL: Add attribute Act.title


O.K. let's look at the definition again carefully:

<definition>
A word or phrase by which an Act may be known among people.

Example: name of a research study (e.g. "Scandinavian Simvastatin Study"), name of a court case (e.g. "Brown v. Board of Education"), name of another kind of work project or operation. For acts representing documents, this is the title of the document.

Discussion: This is not a formal identifier but rather a human-recognizable common name. </definition>

So, what is a title? A "symbol for something", well, apart from the fact that words are symbols in the broader sense, it is explicitly a name known to people but not a formal identifier. We say that quite clearly.

Can it be equally represented ...? This is free text string and only that. It is NOT a code or id. Hence the only way to represent it is as a string.

Is there a rule that says a title cannot be represented in
id, classCode, code or elsewhere? Not one rule, but there is no rule that says title can be representen in those either. As I said earlier, the definition of originalText doesn't really lend itself for titles in general, because titles are not necessarily the basis for coding the Act.code.

So, if there is something that you think should go into the definition, please say what it might be.

thanks,
-Gunther

Russler, Dan wrote:
> I hope what I have written is not perceived as resistance as much as 
> trying to understand and then make suggestions for fine tuning the
proposal.
>
> My sense is that the proposal is still fuzzy. It may be that the 
> proposal needs to be fuzzy. Are we as good as we can get?
>
> To ask the question, what is a title?...Is it a symbol for 
> something?...If it is a symbol, can it equally be represented by 
> different methodologies for symbols: free text strings, codes, icons, 
> etc....What symbols can the datatype cd represent?...Are all these 
> symbols constrained in the definition to specific uses. Is there a 
> specific constraint that says a title cannot be represented in one of 
> the symbols?
>
> If you can differentiate title in an unambiguous way from all the 
> symbol fields in cd, then we should have a separate title attribute. 
> I'm not sure you can do that.
>
> If you can't differentiate title in an unambiguous way, we still can 
> have a separate title attribute, but we have to me more careful on how 
> we guide users to use the field. My sense is just that we have not yet 
> given enough guidance in the definition and examples. I'm not sure 
> that I am resistant to the idea as a whole, just the fact that the 
> definition and examples are still not enough guidance.
>
> It could be that some of what you say below needs to go into the 
> act.code definition and examples as part of this proposal.
>
> -----Original Message-----
> From: Gunther Schadow [mailto:gunther@aurora.regenstrief.org]
> Sent: Wednesday, July 09, 2003 2:06 PM
> To: Russler, Dan
> Cc: 'QUADE_LINDA@LILLY.COM'; mnm@lists.hl7.org; Bob Dolin; Sandy Boyer
> Subject: Re: RIM PROPOSAL: Add attribute Act.title
>
>
> Hi Dan,
>
> I am quite surprized about the level of resistance. You might have a 
> point that I am not getting. Clearly the examples are not suggesting 
> any particular codes, just what could be codes and how the codes do 
> not derive literally from the titles (unless you know what the titles 
> stand
> for.) Because "Shared Pathology Information Network" could be a
> commercial project, a funded research project, a protocol description,
> or just a proposal document, it is not really originalText to any sort
> of Act.code. The Act.code never encodes this title but codes the kind of
> Act that is known from elsewhere, but not from the title.
>
> I am surprized you argue with "levels of granularity" because most 
> people would say that a class of acts is something qualitatively 
> different from an instance of an act. Of course such distinction is 
> somewhat naive and the RIM with the moods crosses this boundary in 
> many interesting ways. So, you have a point that an Act.code really 
> could be a "finely granular" as to label only a single Act instance. 
> But this is probably very hard to sell to people who just think they 
> know the difference between a class and an instance. That's why the 
> Act.title will relax the need to take up this somewhat difficult 
> discussion and make it just easier to work with.
>
> Act.code.originalText will then be restored to what it really wants to 
> be, i.e., a place to put the original text that was the basis for the 
> coding. If the code was assigned by other means than post-coding text, 
> then originalText is null. This is very simple. Since a title can 
> exist even if the Act.code is pre-coded and not derived from any other 
> text, it has always been incorrect to use Act.code.originalText for a 
> title.
>
> So, the discussion should be on much simpler grounds. The reason why 
> everyone must be in favor of this proposal is that the Act 
> .code.originalText is not an appropriate place to put titles that are 
> not the basis for post-coding of Act.code. Even if Act.code is 
> post-coded from a title, the originalText that was the basis for 
> coding may not be the full title but may only be part of it. 
> Therefore, originalText is not a good place to expect the full title 
> to always be. Hence there needs to be a place where you can count on 
> that the full title will be (if it exists), and that purpose has not 
> been served by anything, and that's why we need the Act.title 
> attribute.
>
> regards,
> -Gunther
>
> Russler, Dan wrote:
>  > I am starting to feel better about the proposal, although what the  
> > examples show is purely a classification system demonstrating levels
of
>  > granularity which need to be enforced by custom rather than logical  
> > rules related to instances of classes; just like the difference
between
>  > act.code and act.text. The proposal can still be workable as long 
> as  > custom dictates the level of granularity found in the free text 
> field  > attributes of (act act.code; act.title; act.text attributes)
acceptable
>  > for a given domain.
>  >
>  > I'd support it if there were some way to approach these three free  
> > text  > areas we would have with a set of rules for guidance to 
> future  > classifiers of given domains...The examples help...that is 
> assuming
that
>  > there is agreement on the codes for "medication clinical trial,"
"court
>  > case," etc...In the examples below, medication clinical trial seems 
> to  > be a different level of granularity than "NIH funded research"
---simply
>  > because NIH funded research could fund a medication clinical
trial...But
>  > with discipline, we could eliminate those problems...The real 
> problem
is
>  > that these examples aren't quite sufficient or self-explanatory 
> enough  > for an adequate definition...Dan  >
>  > -----Original Message-----
>  > From: Gunther Schadow [mailto:gunther@aurora.regenstrief.org]
>  > Sent: Wednesday, July 09, 2003 11:10 AM
>  > To: Russler, Dan
>  > Cc: 'QUADE_LINDA@LILLY.COM'; mnm@lists.hl7.org; Bob Dolin; Sandy Boyer
>  > Subject: Re: RIM PROPOSAL: Add attribute Act.title
>  >
>  >
>  > I think we gave enough examples for why Act.title and Act.code can be
>  > different and why Act.title makes sense. Again, examples are:
>  >
>  > title                                  code
>  > -------------------------------------- -----------------------------
>  > 3rd Lilly Prozac Study                 medication clinical trial
>  > Scandiavian Simvastatin Study          medication clinical trial
>  > Indianapolis Network for Patient Care  NIH funded research project
>  > Shared Pathology Information Network   NCI funded research project
>  > Brown vs. Board of Education           Court case
>  > Health for all by the year 2000        WHO program
>  > Guidelines for Risk Stratification     Clinical Guideline
>  >     after Myocardial Infarction
>  >
>  > in all these cases the code doesn't necessarily derive from the title,
>  > nor is the title derived from the code. Also in these cases the title
>  > isn't obviously something that specifies the "kind of Act".
>  >
>  > So, I don't hear an argument against the title. Notice that we have
>  > considered something like Act.name on and off over the years. The only
>  > reason I think we never really pursued it until now is that we
>  > considered Act mostly as a healthcare service event, and there it is
>  > exceedingly rare that someone gives an individual title to a
healthcare
>  > service event (like "Ole Ollies Odd Omentum Operation" doesn't 
> happen
so
>  > often.)
>  >
>  > As far as its relation to mood goes, we just have to consider what  
> > makes  > most sense: clearly we use Act.code in definition mood and 
> event mood
as
>  > a classifier for a kind; but because definition mood Acts are  > 
> descriptions of kinds of acts by themselves, an ActDefinition.code may  
> > feel like the name or title. However, ActEvent.code is for most 
> people  > understandible as only a code for the kind of event, not for 
> the  > individual occurrence. Conversely, with Act.title, people will  
> > understand that Act.title is free to choose differently for each  > 
> individual event, although there is no requirement that title be
unique
>  > between two act events. Conversely, Act.title may feel overlapping
with
>  > Act.code in the case of definition mood, but as we can see above, 
> that  > is not necessarily the case.  >
>  > So, let us have Act.title for anyting that humans consider the name or
>  > title of an act and that they don't want to consider a code. Make sure
>  > that we never turn the Act.title into a "string with code" and that we
>  > understand that the only true handle on an Act instance for a computer
>  > is its Act.id, and not its title. If we just use these common sense
>  > considerations, we'll be safe; in fact safer than we used to be when
we
>  > told people to use Act.code for everything.
>  >
>  > regards,
>  > -Gunther
>  >
>  > Russler, Dan wrote:
>  >  > I don't think there is any reason that the information within 
> the  > cd  > datatype can't evolve over the lifecycle of an 
> instance...For  > example, a  > text string placed in original text 
> may be coded by one  > coder to SNOMED  > as an update to the 
> instance; later, another coder  > may add the ICD code  > to the 
> instance; later, another person may add  > a Spanish text string to  > 
> the instance so that the meaning can be  > printed in Spanish; none of  
> > these activities are necessarily  > perfect one to one mappings, but  
> > reflect the kinds of activities  > done every day to instances of 
> information.  >  >  > So I think we need to be specific on how 
> act.title differs from  >  > act.code  >  > on how it modifies the 
> meaning of the instance. The idea that both  >  > act.code and 
> act.title could both be in definition mood is  >  > troublesome...it 
> would mean that two instances in event mood could have
>  >  > both the same act.title and act.code...If that is so, then there
> is much
>  >  > overlap between the text strings in act.code and the text string in
>  >  > act.title....Dan
>  >  >
>  >  > -----Original Message-----
>  >  > From: QUADE_LINDA@LILLY.COM [mailto:QUADE_LINDA@LILLY.COM]
>  >  > Sent: Tuesday, July 08, 2003 5:24 PM
>  >  > To: Gunther Schadow
>  >  > Cc: Russler, Dan; mnm@lists.hl7.org; Bob Dolin; Sandy Boyer
>  >  > Subject: Re: RIM PROPOSAL: Add attribute Act.title
>  >  >
>  >  >
>  >  > using act.code original text for the title also made it impossible
to
>  >  > use act.code for any legitimate classification purposes. Linda  
> >  > _________________________  >  > Linda M. Quade
>  >  > Associate Information Consultant
>  >  > Eli Lilly & Company
>  >  > Lilly Corporate Center, DC 3047
>  >  > Indianapolis  IN  46285
>  >  >
>  >  > LindaQ@Lilly.com
>  >  > phone: 317.276.9874
>  >  > fax: 317.277.8768
>  >  >
>  >  >
>  >  >
>  >  >
>  >  > Gunther Schadow <gunther@aurora.regenstrief.org>
>  >  > 07/08/2003 03:59 PM
>  >  >
>  >  >
>  >  >         To:     "Russler, Dan" <Dan.Russler@McKesson.com>
>  >  >         cc:     mnm@lists.hl7.org, Linda Quade
> <QUADE_LINDA@LILLY.COM>,
>  >  > Bob Dolin
>  >  > <Robert.H.Dolin@kp.org>, Sandy Boyer <slboyer@attglobal.net>
>  >  >         Subject:        Re: RIM PROPOSAL: Add attribute Act.title
>  >  >
>  >  >
>  >  > The two fields do not do the same thing. There is always ambiguity
>  >  > left
>  >  > with Act.code.orginalText and it isn't always appropriate to use it
>  > at all.
>  >  >
>  >  > Original text is used when something is encoded after it had been
>  >  > written down. If you author an Act directly and pull a code from a
> list,
>  >  > then you don't have original text, but you still might have a
title.
>  >  >
>  >  > Act.code is believed to represent the kind of act whereas an Act 
> title  >  > is supposed to name an instance. The difference is blurry, 
> but if
you
>  >  > need a title such as "Brown v. Board of Education" then this is 
> not the  >  > original text for a code "SupremeCourtDecisionAct".
>  >  >
>  >  > Act title is in many cases more akin to something in the Act.id
than
>  >  > in
>  >  > the Act.code.
>  >  >
>  >  > The practice to use Act.code.orginalText was a crutch at best 
> and
has
>  >  > been very confusing and hard to justify upon closer look.  >  >
>  >  > regards
>  >  > -Gunther
>  >  >
>  >  >
>  >  > Russler, Dan wrote:
>  >  >  > So do we need better education or a new attribute?...Do you
> want to
>  >  > > make
>  >  >
>  >  >  > the original text an optional field?...My main point was that we
>  >  > would  > now have two fields for the same thing...The proposal
> adds an
>  >  > attribute,
>  >  >
>  >  >  > but didn't take anything away from act.code...Dan
>  >  >  >
>  >  >  > -----Original Message-----
>  >  >  > From: Gunther Schadow [mailto:gunther@aurora.regenstrief.org]
>  >  >  > Sent: Tuesday, July 08, 2003 4:40 PM
>  >  >  > To: Russler, Dan
>  >  >  > Cc: mnm@lists.hl7.org; Linda Quade; Bob Dolin; Sandy Boyer  >
>  >  > Subject: Re: RIM PROPOSAL: Add attribute Act.title  >
>  >  >  >
>  >  >  > As I said in the rationale: yes, it was in
> Act.code.originalText, but
>  >  >  > it
>  >  >
>  >  >  > was never quite clear there
>  >  >  >
>  >  >  > Russler, Dan wrote:
>  >  >  >  > Up to now, I believe this act title would be found in the
>  >  > act.code
>  >  >  > > within the cd datatype as the text description field...or am I
>  >  > missing
>  >  >  >  > the point?...Dan
>  >  >  >  >
>  >  >  >  > -----Original Message-----
>  >  >  >  > From: Gunther Schadow [mailto:gunther@aurora.regenstrief.org]
>  >  >  >  > Sent: Tuesday, July 08, 2003 4:33 PM
>  >  >  >  > To: mnm@lists.hl7.org
>  >  >  >  > Cc: Linda Quade; Bob Dolin; Sandy Boyer
>  >  >  >  > Subject: RIM PROPOSAL: Add attribute Act.title
>  >  >  >  >
>  >  >  >  >
>  >  >  >  > Hi, Abdul-Malik said that we could issue this proposal
> through the
>  >  >  > > email  > list without a cover sheet and spread sheet.
>  >  >  >  >
>  >  >  >  > ADD ATTRIBUTE Act.title : ST
>  >  >  >  >
>  >  >  >  > <definition>
>  >  >  >  > A word or phrase by which an Act may be known among people.
>  >  >  >  >
>  >  >  >  > Example: name of a research study (e.g. "Scandinavian
> Simvastatin
>  >  >  >  > Study"), name of a court case (e.g. "Brown v. Board of
>  > Education"),
>  >  > name
>  >  >  >  > of another kind of work project or operation. For acts
>  > representing
>  >  >  > > documents, this is the title of the document.  >
>  >  >  >  > Discussion: This is not a formal identifier but rather a
>  >  >  >  > human-recognizable common name. </definition>
>  >  >  >  >
>  >  >  >  > This proposal is being put forth jointly by the Technical
>  > Committees
>  >  >  >  > RCRIM and ORD and SDTC.
>  >  >  >  >
>  >  >  >  > Rationale: in the past one had to use Act.code.originalText
to
>  > place
>  >  > a
>  >  >  >  > document title and it wasn't clear whether a given  >  > 
> Act.code.orginalText  >  >  >  > really is supposed to be a title. Use 
> case for titles is  >  >  > (obviously)  > document titles but also 
> project names such as  > required
>  >  >  > for research  > studies.  >
>  >  >  >  > thanks for the consideration,
>  >  >  >  > -Gunther
>  >  >  >  >
>  >  >  >  > --
>  >  >  >  > Gunther Schadow, M.D., Ph.D. gschadow@regenstrief.org
>  >  >  >  > Medical Information Scientist      Regenstrief Institute for
>  > Health
>  >  > Care
>  >  >  >  > Adjunct Assistant Professor        Indiana University School
of
>  >  > Medicine
>  >  >  >  > tel:1(317)630-7960
>  >  > http://aurora.regenstrief.org
>  >  >  >  >
>  >  >  >  >
>  >  >  >  >
>  >  >  >  > ---
>  >  >  >  > To access the Mailing List archives, go to:
>  >  >  >  > http://lists.hl7.org/lyris.pl?enter=mnm
>  >  >  >  >
>  >  >  >
>  >  >  >
>  >  >  > --
>  >  >  > Gunther Schadow, M.D., Ph.D.
>  > gschadow@regenstrief.org
>  >  >  > Medical Information Scientist      Regenstrief Institute for
>  > Health Care
>  >  >  > Adjunct Assistant Professor        Indiana University School of
>  > Medicine
>  >  >  > tel:1(317)630-7960
>  > http://aurora.regenstrief.org
>  >  >  >
>  >  >
>  >  >
>  >  > --
>  >  > Gunther Schadow, M.D., Ph.D.
> gschadow@regenstrief.org
>  >  > Medical Information Scientist      Regenstrief Institute for
> Health Care
>  >  > Adjunct Assistant Professor        Indiana University School of
> Medicine
>  >  > tel:1(317)630-7960
> http://aurora.regenstrief.org
>  >  >
>  >  >
>  >  >
>  >  >
>  >  >
>  >  >
>  >  > ---
>  >  > To access the Mailing List archives, go to:
>  >  > http://lists.hl7.org/lyris.pl?enter=mnm
>  >  >
>  >
>  >
>  > --
>  > Gunther Schadow, M.D., Ph.D.
gschadow@regenstrief.org
>  > Medical Information Scientist      Regenstrief Institute for Health
Care
>  > Adjunct Assistant Professor        Indiana University School of
Medicine
>  > tel:1(317)630-7960
http://aurora.regenstrief.org
>  >
>  >
>  >
>  > ---
>  > To access the Mailing List archives, go to:
>  > http://lists.hl7.org/lyris.pl?enter=mnm
>  >
>
>
> --
> Gunther Schadow, M.D., Ph.D.                    gschadow@regenstrief.org
> Medical Information Scientist      Regenstrief Institute for Health Care
> Adjunct Assistant Professor        Indiana University School of Medicine
> tel:1(317)630-7960                         http://aurora.regenstrief.org
>


--
Gunther Schadow, M.D., Ph.D.                    gschadow@regenstrief.org
Medical Information Scientist      Regenstrief Institute for Health Care
Adjunct Assistant Professor        Indiana University School of Medicine
tel:1(317)630-7960                         http://aurora.regenstrief.org



---
To access the Mailing List archives, go to: http://lists.hl7.org/lyris.pl?enter=mnm







---
To access the Mailing List archives, go to: 
http://lists.hl7.org/lyris.pl?enter=mnm

