[Top] [All Lists]

[health-ont] Comments on Time concepts in HL7

To: <health-ont@xxxxxxxxxxxxxxxx>
From: "Patrick Cassidy" <pcassidy@xxxxxxxxx>
Date: Thu, 10 Feb 2005 13:47:58 -0500
Message-id: <200502101848.j1AIm6U11846@xxxxxxxxxxxxxxxxxxxxxxxxx>
An initial attempt was made to review time-related concepts in HL7 to
determine how they might map to concepts in an upper ontology.
Some initial comments are attached as a WinWord file and also appended as
text.  There are additional time-related concepts in HL7, and in particular
relations of time to other concepts.  These will be reviewed as time
permits, or if others request further discussion.    (01)

  Pat    (02)

Patrick Cassidy
MITRE Corporation
W903 Information Discovery & Understanding
7515 Colshire Drive
McLean, VA 22102-7508
Phone: 703-883-8976
(Eatontown: 732-578-6340)
Fax: 703-883-1379
Email: pcassidy@xxxxxxxxx
Mail Stop: MNJE    (03)

=====================================    (04)

Considerations for aligning HL7 Time concepts with an upper ontology    (05)

As part of the Ontolog group participation in discussion of the National
Health Information Network RFI, we are providing information on how
ontologies can assist in enabling the interoperability of patient health
records.  At this point, we are focused on how existing and prospective
health domain ontologies and taxonomies can be aligned with upper ontologies
to improve the accuracy of conceptual information transfer among systems
using different domain knowledge representations.    (06)

For this facet of that task, my immediate focus will be to determine how the
representations of time in the HL7 terminology may be related to similar
concepts in an upper ontology.    (07)

HL7 Information Request    (08)

I have looked at the lists of data types in the html files of the current
HL7 Reference Model.  I also found some definitions of the types in the XML
files.  However, I have found only a few files related to the relations
between concepts.  I would appreciate help from anyone who can point me to
the locations of a comprehensive list of the relations that are defined
between concepts within the HL7 system.  In particular, I can't find
anything describing the relations between time intervals, time points, or
events that may be specified as taking place at a particular time.    (09)

There are a few additional questions I have:    (010)

Regarding the term "DataType": is this intended to mean that the Type labels
used here may be attached to field designators in documents, such as in
XML-tagged documents?  Or are these intended merely as general terms for
concepts?  Or both?    (011)

DataValue: can a data value be qualitative (e.g. color)?  I could not find
color in the classification.  Where is that located?    (012)

In examining the files I can find in HL7, there is one issue regarding the
classification of Time data types that can be discussed with the information
I currently have.  That issue is discussed below.  If anyone has additional
information relating to there concepts in HL7, or if I have misinterpreted
the intended meanings of these concepts, please send me a note
(cassidy@xxxxxxxxx) or send a note to this list.    (013)

Time Representation
General comments on time representation:
   In order to represent time so as to be broadly compatible with many
different databases, it is necessary to make at least the following
distinctions:    (014)

  (1) Specific calendar times such as 12:00 Noon GMT on Feb. 5, 2005 need to
be distinguished from a specific quantity of time, such as 20 minutes.
 (2) A relation between time points of before and after needs to be defined.
(3) Relations between specific time intervals, such as coinciding,
overlapping, or before/after, need to be defined.
(4) Recurring time intervals, such as days of the week or months of the
year, need to be distinguished from time points, specific time intervals,
and quantities of time.    (015)

Time Representation in HL7    (016)

Within the HL7 data types the following (selected) portion of the taxonomy
was noted:    (017)

      DataTypePeriodicIntervalOfTime    (018)

   DataTypePointInTime    (019)

It appears that DataTypePointInTime is intended to refer to a specific point
in time on the calendar - a specific time location.
>> A quantity specifying a point on the axis of natural time. A point in
time is most often represented as a calendar expression    (020)

  An interval of time points (DataTypeIntervalOfPointsInTime) would then
also be a specific time location.    (021)

DataTypeGeneralTimingSpecification  is defined as:
>> A set of points in time, specifying the timing of events and actions
and the cyclical validity-patterns that may exist for certain kinds of
information, such as phone numbers (evening, daytime), addresses (so
called "snowbirds," residing in the south during winter and north
during summer) and office hours.    (022)

There is a problem with this arrangement:  The instances of the parent class
(Type) DataTypeIntervalOfPointsInTime are each a single interval of time,
but the suggested instances of this class (evening, wintertime), are
conceptually each themselves not single time intervals, rather they are
classes of time intervals (i.e., all evenings, etc.).  For example, in the
SUMO ontology, Monday is a subclass of TimeInterval, not an instance.  Thus
this class (Type) would not fit into a schema where instances are intended
to be individuals. This schema above can work for its intended purpose if
users did not attempt to draw logical inferences from the subclass structure
that is given here, but merely used the terms in a single restricted manner.
      If this conceptual classification is intended to at some point map to
other classifications, however, it will be important to keep  the logical
meanings consistent for all inherited instances within a subclass structure.
In this case, there are two possibilities to modify the existing
(1)  the class (Type) DataTypeGeneralTimingSpecification can be left in its
present position, but the types of recurrent time interval which are
intended to be used (evening, Winter, Mondays from 9 to 5) should each be
entered as a subclass (subtype) of that class, rather than as instances.
Such subtypes will themselves have instances which are specific time
intervals  (e.g., Monday, Feb. 7, 20005, GMT would be an instance of the
class Monday), but perhaps more commonly the classes (e.g. Monday) would be
the individual data elements that are used in medical records or other
(2) If it is desired that all data elements used in data communications be
instances of a designated data type, and must not be  subclasses, the class
of  DataTypeGeneralTimingSpecification should be moved from its present
position to a higher level in the DataType hierarchy, at the same level as
DataTypeIntervalOfPointsInTime.  Then each instance of this class , though
it is also conceptually a class, could be used as a data value while
conforming both to the general schema, and to a strict logical inheritance
rule that will enhance interoperability.    (023)

On the related Type DataTypePeriodicIntervalOfTime, I could not find the
notes describing its intended meaning, and it is not clear how this differs
from DataTypeGeneralTimingSpecification.  Perhaps this is intended to
represent strictly periodic intervals?  or blocks of time, such as one week?
The definition or a pointer to the definition of this concept would be
helpful.    (024)

Attachment: HL7time.doc
Description: MS-Word document

Msg Archives:
Community Files:
To  Post: mailto:health-ont@xxxxxxxxxxxxxxxx
Community Wiki:    (01)
<Prev in Thread] Current Thread [Next in Thread>