Arden Syntax SIG
Meeting Minutes
Winter Working Group Meeting, San Diego
January 8, 2002
Author:  Robert Jenders


Attendance:  Pat Lyons (Siemens; pat.lyons@smed.com), Matt Sailors (UT-Houston; matthew.sailors@uth.tmc.edu), Peter Haug (LDS Hospital; LDPHaug@ihc.com), Sven Tiffe (Siemens; sven.tiffe@med.siemens.de), Robert Jenders (Columbia U; jenders@columbia.edu), John Dulcey (Applied Medical Informatics; john.dulcey@alum.mit.edu), Tom Hooks (McKesson; tom.hooks@mckesson.com), Leilei Wang (Epic Systems; leilei@epicsystems.com), Sarah Ryan (CDC; sryan23@csc.com), Harm Scherpbier (Siemens; harm.scherpbier@smed.com).

1.  Arden v2.1:  New Operators (Hooks; McKesson)
        a.  currenttime:  System clock time at the moment of execution (as opposed to "now",
                which refers to system/clock time at time MLM begins execution):  Approved 10 yes,
                0 no.
        b.  length:  Returns length of (string) operator.
                1.  Discussed problem of length of a number:  May be system-dependent (precision of
                        numbers; casting).
                2.  Approved 10 yes, 0 no, with clarification to make distinction with count of list.
        c.  uppercase/lowercase:  Swap one with the other, if possible.  Approved 10 yes, 0 no.
        d.  trim:  remove leading and trailing white space.  Approved 10 yes, 0 no.
        e.  find...[in] string [starting at]:  locate starting index of literal pattern in a string.
                Approved 9 yes, 0 no.
        f.  extract characters...through...from
                1.  renamed to substring because of extant "extract characters" operator.
                2.  new version:  substring <n:number> characters [starting at <n:number>] from
                        <n:string>
                        a.  if # characters is negative, returns characters going left from start point
                        b.  "starting at" is optional; if omitted, assumes 1 (beginning of string).
                        c.  First character returned is that at index position
                        d.  Preserve primary time.
        g.  Action item:  Write in format of standard [Hooks] and compose BNF [Sailors].

2.  v2.1 Process
        a.  Action item:  Include items (new operators above, XML changes) from this meeting [Sailors].
                1.  Goal deadline for submission to Sailors:  February 1.
                2.  Goal deadline for distribution of assembled document:  February 4.
        b.  Action item:  Detailed reading and fixing bugs [Sailors, Jenders, Dulcey, Siemens TBN by Lyons].
                1.  May include listserve iterations if substantive problems discovered.
                2.  Goal deadline for initial review:  February 18.
                3.  Goal deadine for distribution of revision:  February 22.
                4.  Goal deadline for second review:  March 4.
                3.  Goal deadline for completion of final document:  March 8.
                4.  Discuss ballot at ATL meeting in late April, 2002.

3.  Outreach Efforts
        a.  Implementation guide:  Outline pending.
                1.  Action item:  Complete outline and circulate [Sailors].
                2.  Discussion:  Use format of HL7 Implementation Guide as basis.
        b.  Structuring of outreach presentations:  IEEE standard LOM = Learning
                Object Model (Dulcey brief presentation)
                1.  Variations:  DOD version:  ADL/SCORM, IMS Global Consortium version
                2.  Contains units ("bits") of knowledge that would be 1 or a small group
                        of slides regarding a specific idea.
                3.  Apply metadata (level of difficulty; topic; audience).
                4.  Put in shareable library.
                5.  Action item:  Discuss with Education Committee [Dulcey].

4.  Reconciling CDA with Arden v2.1 suggested XML representation:  Joint Meeting with
        Structured Documents TC (represented by Bob Dolin, TC co-chair)
        a.  Sailors reviewed current levels 0-4 for Arden.
        b.  To adopt CDA, MLM constituents must be mapped to the RIM (with appropriate
                type codes), followed by R-MIM development and then schema creation.
        c.  Discussion:  No compelling need for XML representation independent of the
                RIM now.  Mapping to the RIM will require more time than is available for Arden
                v2.1.  Therefore, consensus is to defer this mapping to Arden v3.
        d.  Decision (11 yes, 0 no):  Remove suggested XML representation from the
                proposed v2.1 and rewrite as a separate, informative document, reducing from
                5 to 3 levels.
        e.  Timeline for this representation:  Circulate in advance of and review at Fall, 2002
                meeting.

5.  Discussion of Proposed Common Expression Language
        a.  One concern expressed:  Decreased readability may lead to less vendor
                customer acceptance.
        b.  Consensus:  Object-oriented (dot notation) data model (without Arden operators
                mapped to methods as in the current version of the GELLO expression language) using
                vMR will help solve the curly-braces problem while maintaining legibility.
                1.  Problem = methods may be essential in GLIF/guideline formalism.
                2.  Constitutes overloading of current Arden operators.
        c.  Variations:
                1.  Object-oriented queries but keep current Arden data model and operators,
                        "flattening out" database objects.
                            a.  Advantage:  Minimal change to current Arden.
                            b.  Disadvantage:  Objects become collections of parallel lists.
                2.  #1, but add a few attributes to the current model (value and primary time).
                            a.  Advantage:  Some change for current Arden operators required, but not much.
                            b.  Disadvantage:  Still need to "flatten" complex (vMR/database) objects.
        d.  Problem = backward compatibility and lack of vMR mappings at a particular
                site.  Consensus:  Allow use of curly braces to allow querying at sites without a vMR-
                compliant system.
        e.  Problem:  GELLO uses stronger typing not currently in Arden.  Consensus:  Avoid this.
        f.  Overall consensus:  Object references (vMR dot notation without methods) on the read
                RHS.  Undecided (to be decided through further discussion):  keep read LHS as
                simple Arden variables vs slightly more complexvariables (matrices) vs
                object-oriented dot notation.

6.  Arden v3 issue:  Further discussion of fuzzy logic
        a.  Sven Tiffe reviewed further work on this topic (project TOSCA for diagnosing glaucoma).
                1.  linguistic variable:  variable with possible fuzzy values, e.g., "normal", "increased".
                2.  each value then is defined as a fuzzy set, requiring a new definition for every different
                        use.
        b.  Ultimately deferred to Arden v3.
 

January 11, 2002  0900-1230

Attendance:  Matt Sailors (UT-Houston; matthew.sailors@uth.tmc.edu), Sven Tiffe (Siemens; sven.tiffe@med.siemens.de), Robert Jenders (Columbia U; jenders@columbia.edu), Tom Hooks (McKesson; tom.hooks@mckesson.com), John Dulcey (Applied Medical Informatics; john.dulcey@alum.mit.edu),  Lei Lei Wang (Epic; lwang@epicsystems.com), Peter Haug (LDS Hospital; LDPHaug@ihc.com), Robert Hausam (TheraDoc; robert.hausam@theradoc.com), Stan Huff (Intermountain Health Care;).

1.  Arden Syntax v3:  Fuzzy Logic [Sven Tiffe presentation]
        a.  linguistic variables:  method for defining fuzzy functions for each context (e.g.,
                "normal weight", "high blood pressure").
        b.  a categorical value (e.g., the actual blood pressure) then is mapped to one value
                (e.g., "high") of the linguistic variable.
        c.  Proposes a separate MLM to map some input value to a linguistic variable,
                necessitating new slots in the knowledge category of the MLM.
        d.  <linguistic var> IS <some linguistic variable value> => returns a fuzzy truth value.
        e.  Accommodating fuzzy logic (and, generally, other measures of uncertainty) were
                discussed.  Consensus:  Continue work in anticipation of using this in Arden v3.
        f.  Action item:  Summarize/document current work and circulate in advance of Atlanta
                meeting [Tiffe].

2.  Arden as XML
        a.  Discussed utility of various levels of markup.
        b.  Action item:  Work on suggested XML representation as separate informative
                document [Sailors, Dulcey].
 

Agenda for Atlanta Meeting (April 29 - May 3, 2002)
Requested meeting time:  Tuesday, April 30 0900-1700.

1.  Review anticipated ballot for Arden Syntax v2.1.
2.  Outreach:  Review progress on creation of a library of presentations.
3.  Work on implementation guide.
4.  Review work on a common expression language and data model.
5.  Review plan for creation of informative document specifying an XML form for MLMs.