HL7 Version 3 Data Type Redesign Project

The HL7 control query committee on its September 1998 meeting in San Diego set up a task force to work out a redesign proposal for HL7 version 3 data types.

After 13 conference calls that produced lots of discussion and lots of notes, we delivered a report to the HL7 Winter meeting in Orlando 1999. That report was presented to the HL7 Technical Steering Committee (all HL7 co-chairs), was presented in more detail at the Control/Query meeting, and spare copies were distributed randomly to the HL7 meeting attendees.

An estimated 80% of the work was completed, and we were going to do the next 20% until the April 1999 meeting in Toronto.

After Toronto we had 90% complete and a pause began. In San Diego 2000 the importance of this material being balloted became evident. It also became evident that the Report was not suitable as a balloted standard document. So we had to rewrite and complete everything. This work is done now for Cleveland, Spring 2000.

After Cleveland 2000 we received very inspiring pre-ballot comments which caused us to make some significant changes. Now, finally the XML ITS is getting ready and the ballot should start in this first week of August 2000.

History: for those who are interested in the historical evolving of ideas, please refer to the old index page that reflects the status quo of January '99.

HL7 Version 3.0 Data Type Specification

Ballot draft #2.

Here is ballot draft #2 as it is now in HL7 committee ballot. This is now a two-part document as we were requested to provide an introductory abridged and simplified version. There are a few noteworthy changes to the actual content.

• Rearranged the Section-Outline (of Part II, the unabridged specification.)
  • People didn't think the partitioning into Text, Things, and Quantities was appropriate for a standard. So the 14+1 basic data types are now flattened all in on section.
  • Exposition of the Formal Definition Language is now a major part of Section 1 (Introduction) rather than a Section of its own.
  • What was called "Fundamental Data Types" before are now part of the Intro section.
  • Rearrangements at the end, so that timing specification is the last section (5)
  • Added Tables of "primary properties" which is somewhat fuzzy defined. It allows people who otherwise get all confused to see a table of "components" (even though this specification doesn't define a rigid set of components.)
  • Put much of the "rationale" text into footnotes, where this material is not "in the face" of the uninterested (or easily confused) reader. Balloters had suggested to put this into an appendix -- yet, that would have separated each commentary from its subject and would instead have required much more pages of writing just for rationale. The footnotes should be a good compromise (some people are footnote-readers and some aren't.)
• Introduction
  • Condensing the first subsections. Some people found them too laborious ... however, they were laborious because people would always ask the same questions. It didn't help anyway, since people prefer asking questions rather than to read the answers. This into material reads now like a FAQ and hopefully will be read by those who worry about these questions
• Basic Types
  • All basic 14 types listed in this section. The abstract QTY type is also listed there.
  • I was ordered to rename "Encoded Data" to "Encapsulated Data". Sigh, "encapsulation" is an implementation concept :-(
  • A new CD restriction "CC" added in the last minute based on Harmonization meeting Fall 2000.
  • Person Name Type and Organization Name merged into Entity Name, plus three restrictions for Person Name, Organization Name, and Trivial Name.
  • Some cleanup in the INT and REAL axioms.
• Generic Collections
  • Set: now exposes SET<QTY> and adds properties to it: totallyOrdered:BL and hull:IVL<TS> (see below about timing specification.)
  • Interval: added the interval hull operation.
  • Clarified the promotion of TS to IVL<TS>.
  • Literal form of IVL<TS> is changed. Dash-form is now forbidden for IVL<TS> because of conflicts with the timezone notation and the overall quarrel with the W3C's ruling to only allow the adorned ISO 8601 form.
  • Literal form for TS adds a "hull-form" instead of the "dash-form". This actually makes the behavior of TS "ranges" more intuitive (i.e., we would consider 19870219-19870225 as including the whole day of the 25th, but the dash-form didn't say that. The hull form 19870219..19870225 however does mean this.)
• Generic Type Extensions
  • Removed the Annotation (ANT) generic type extension.
  • Removed the Uncertain Value - Narrative (UVN) generic type extension.
  • Imposed a "self-restraint" which forbids generic extensions from being substituted for the basic types except where specifically permitted.
• Timing Specification
  • A thorrough revision of examples showed some inconsistencies especially with the dash-form notation (which is now disabled as explained above.)
  • A missing property of PIVL has been added ("institutionSpecifiedTime:BL)
  • GTS.outerBound:IVL<TS> has been renamed to "hull", since "convex hull" is what this construct is called in the literature.
  • The nextAfter property has been revised the formal definition wasn't right. Also, the conversion between GTS and LIST<IVL<TS>> has been defined now.
  • The concept of "interleaving schedules" and "periodic hull" has been developed. This concept was missing all along and not knowing about it caused a lot of confusion. A periodic hull allows you to say J1..5 for "monday to friday" or more complex periodic intervals that are constructed as the hull of the two "neighboring" occurrence intervals. This now allows you to specify e.g., "Memorial Day to Labor Day" or "the last week of May" (which is impossible to do without the periodic hull.)
  • Parentheses have been added for the GTS expression as Doug Pratt suggested. It does make things easier, even though it's more "stuff".

• PART I: PDF useful for printing, does not contain bookmarks and links.
• PART II: PDF useful for printing, does not contain bookmarks and links.
• ZIPed Word files for those who want to edit it.

OLDER STUFF FOLLOWS

Ballot draft #1.

Responses/Dispositions to Ballot Comments.

Here is ballot draft #1 as it is now in HL7 committee ballot. This document went through some significant changes since Cleveland incorporating suggestions from a number of people who cared to submit their comments early. Thank you!

• PDF for on-screen browsing (contains bookmarks and links that may sometimes interfere with printing.)
• PDF useful for printing, does not contain bookmarks and links.
• ZIPed Word file for those who want to edit it.

The following is really outdated.

Version 0.95 as the basis for discussion.

A new revision is in the pipe. We want to have version 1.0 finished until September, don't we? So, here is new material HTML, an index to the changes, PS, PDF, or ZIP containing the HTML version plus images to download and browse locally.

XML Representation of the types

Here is a proposed XML representation of the data types. This is work that I did in touch with the group, but that does not reflect a consensus. In fact, people are generally sceptical about the "FleXML" and "HyperFleXML" styles of XML use that I am proposing here. If you are curious, please see my presentations to the joint CEN TC251 and HL7 meeting on Wednesday July 25, it's available in PDF (color), PDF (b/w 2up), PowerPoint show (with animations).

Because of my "extremist" views I do not consider myself in charge for the XML mapping any more. Paul Biron and Mark Tucker are probably the ones who'll pick this up.

Version 0.9 as of Toronto 1999

This is the Data Type Specification presented in Toronto: HTML, PS, PDF, or ZIP containing the HTML version plus images to download and browse locally.

This is a stable base upon which we will only have to make modifications and additions. Finally we have the running work document.

Slides

All graphics accessible from here are: Copyright © 1998 Regenstrief Institute. All rights reserved.

First set of slides (San Diego '98)

These were presented before the Control/Query committee.

• From the begining.
• Introduction.
• Text.
• Objects: Concepts and Things.
• Quantities.
• Orthogonal issues.
• Summary, Results, Plan.

A ZIP file of all slides can be downloaded.

based on these we started working ... until we presented in Orlando

Second set of slides (Orlando '99)

These were presented before the Technical Steering Committee.

[view] [download].

Conference Call Notes and Schedule

1. Thursday, October 8, 1998, 11:00 to 12:00 AM EDT.
2. Thursday, October 15, 1998, 11:00 to 12:30 AM EDT.
3. Thursday, October 22, 1998, 11:00 to 12:30 AM EDT.
4. Thursday, October 29, 1998, 11:00 to 12:30 AM EST.
5. Thursday, November 5, 1998, 11:00 to 12:30 AM EDT. (incomplete)
6. Monday, November 23, 1998, 11:00 to 12:30 AM EST.
7. Monday, November 30, 1998, 11:00 to 12:30 AM EST.
8. Monday, December 7, 1998, 11:00 to 12:30 AM EST.
9. Monday, December 14, 1998, 11:00 to 12:30 AM EST.
10. Monday, December 21, 1998, 11:00 AM to 12:30 EST.
11. Monday, December 28, 1998, 11:00 AM EDT. (canceled)
12. Monday, January 4, 1999, 11:00 AM EDT.
13. Monday, January 11, 1999, 11:00 AM EDT.
14. Wednesday, January 13, 1999, 2:00 PM EST. (done, no notes)
15. Monday, February 8, 1999, 11:00 AM EST. (notewriting discontinued)
16. Monday, February 15, 1999, 11:00 AM EST.
17. Monday, February 22, 1999, 11:00 AM EST.
18. Wednesday, February 24, 1999, 3:00 PM EST.
19. Monday, March 1, 1999, 2:00 PM EST.
20. Monday, March 8, 1999, 2:00 PM EST.
21. Monday, March 15, 1999, 4:00 PM EST.
22. Monday, March 22, 1999, 11:00 AM EST.
23. Monday, March 29, 1999, 12:30 AM EST. (done, notes in XML-ITS document)
24. Tuesday, April 6, 1999, 11:30 AM EST. (done, notes in XML-ITS document)
25. Monday, April 12, 1999, 11:00 AM EST. (done, notes in XML-ITS document)
26. Monday, August 2, 1999, 4:30 AM EDT.
27. Monday, August 9, 1999, 10:00 AM EDT. (done, no notes, sorry) Deciding on the small changes and going through issues raised by others.
28. Monday, August 16, 1999, 10:00 AM EDT. (done, no notes, sorry) Periodic points and intervals of time. Intervals as sets, and arbitrary sets of time. Use of periodic intervals in the history collection.
29. Monday, August 23, 1999, 10:00 AM EDT. (done, no notes, sorry) Together with Modelling and Methodology: simplification around TII and RWII. Representation of states.
30. Monday, August 30, 1999, 10:00 AM EDT. (done, no notes, sorry) Concept descriptor and code phrase for post coordinated codes beyond juxtaposition of code values.
31. Tuesday, September 7, 1999, 2:00 PM EDT. (done, no notes, sorry) Concluding on semi-difficult issues (TIL, phone, etc.) XML-ITS 1 relationship between XML Schema and V3DT
32. Monday, September 13, 1999, 5:00 PM EDT. (done) XML-ITS 2 review proposal drafted by Paul Biron and Wes Rishel.
33. Thursday, September 16, 1999, 5:00 PM EDT. (done) XML-ITS 3 continue finalization of Wes' and Paul's proposal.
34. Wednesday, September 22, 1999, 5:00 PM EDT. (scheduled) XML-ITS 3 continue finalization of Wes' and Paul's proposal.

Next conference call is next Wednesday, September 22, 1999, Agenda items for time is (still) to finalize the XML representation of the types.

Please dial 1-800 869 6684 or 1 612 368 3900 to connect. Ask for the HL7 call chaired by Gunther Schadow.