Standards Supporting Medical Communication

1
Standards Supporting Medical Communication
Data Formats and Communication Standards James
H. Harrison, Jr., M.D., Ph.D. Division of
Pathology Informatics Department of
Pathology University of Pittsburgh Medical
Center CLS 5807-MT, 200 Lothrop St. Pittsburgh,
PA 15213 jhrsn_at_pitt.edu
2
The Importance of Standards

• Crucial for widespread acceptance of technology
• Standards vs. innovation
• Standards, interoperability and modularity
• Voluntary or legally mandated
• Independent standards vs. "de facto standards"
• Standards defined by
• Users (HL7, HTML)
• Organizations (IEEE, NEMA, CAP)
• US Government (ANSI, NIST)
• Other governments (CEN)
• United Nations (CCITT, ISO)

3
Standards Supporting Clinical Communication

• Communications
• TCP/IP
• CORBA/DCOM/RMI (communications)
• Data structure and representation
• ASCII, Unicode, RTF, HTML, XML, HL7
• JPEG/MPEG, GIF, TIFF, Quicktime
• DICOM (includes some communications)
• Standard vocabularies/concept lists
• ICD, CPT, SNOMED, LOINC, UMLS, others

4
TCP/IP

• Internet Protocol (IP)
• Responsible for moving packets of data across the
  Internet. IP forwards each packet based on a four
  byte destination address (the IP number)
• Transmission Control Protocol (TCP)
• Responsible for dividing data into packets,
  reassembling the packets, and verifying the
  correct delivery of data from client to server.
  TCP detects errors or lost data and triggers
  retransmission until the data is correctly and
  completely received.

5
ISO Networking Reference Model
File transfer, email, Web Data formatting/compress
ion Authentication/authorization End-to-end
communication Packet routing Ethernet, FDDI,
etc. Fiber, coax, UTP, µwave, RF
Function
Communication
6
Standards for Digital Text

• Every typed character is represented by a number
  (including spaces, tabs and carriage returns)
• American Standard for Computer Information
  Interchange (ASCII) and Unicode
• ASCII
• ASCII is a 128-character set 8 bits (one byte)
  represents one character. Unicode represents
  characters with two bytes (currently 50,000
  char).

Hello there, Nerds!

72 101 108 108 111 32 116 104 101 114 101 44 13
78 101 114 115 33
7
Line Endings in Text Files

• Unix linefeed (ASCII 10)
• DOS carriage return linefeed (ASCII 13 10)
• Mac carriage return (ASCII 13)
• Odd characters may appear when platforms are
  crossed utility software can deal with this

8
Text Formatting in Word Processors
Area of Concentration in Biomedical
Informatics Proposal submitted by James H.
Harrison, Jr., M.D., Ph.D. and Charles Friedman,
Ph.D. University of Pittsburgh Center for
Biomedical Informatics June 16, 2000
Update Overview Biomedical Informatics is a
rapidly growing subspecialty of medicine that is
concerned with biological data analysis and the
management of clinical information to optimize
clinical practice and outcomes. The field covers
a broad range of topics, from artificial
intelligence and automated pattern recognition,
through database structure, information encoding,
management and analysis of biomedical research
information, and user interface design, to system
implementation, outcomes analysis and clinical
process improvement. The evolution of medical
practice and the ongoing dramatic improvements in
hardware and software for information processing
combine to make Biomedical Informatics one of the
most rapidly-developing and exciting areas in
Medicine. Students who have interests and/or
skills in computing and information management
may gain an introduction to Biomedical
Informatics and its application at the University
of Pittsburgh School of Medicine by participating
in a proposed Area of Concentration in Biomedical
Informatics.
9
Text Formatting -- MS Word
gt R
TQ








s Ejbjb
hkk_at_



8Lk
"
(
,UIV

V

T(
JJ
(N
NZ
(
fqo


Area of
Concentration in Biomedical Informatics Proposal
submitted by James H. Harrison, Jr., M.D., Ph.D.
and Charles Friedman, Ph.D. University of
Pittsburgh Center for Biomedical Informatics June
16, 2000 Update Overview Biomedical Informatics
is a rapidly growing subspecialty of medicine
that is concerned with biological data analysis
and the management of clinical information to
optimize clinical practice and outcomes. The
field covers a broad range of topics, from
artificial intelligence and automated pattern
recognition, through database structure,
information encoding, management and analysis of
biomedical research information, and user
interface design, to system implementation,
outcomes analysis and clinical process
improvement. The evolution of medical practice
and the ongoing dramatic improvements in hardware
and software for information processing combine
to make Biomedical Informatics one of the most
rapidly-developing and exciting areas in
Medicine. Students who have interests and/or
skills in computing and information management
may gain an introduction to Biomedical
Informatics and its application at the University
of Pittsburgh School of Medicine by participating
in a proposed Area of Concentration in Biomedical
Informatics. ...

• Multiple copies in the file
• More formatting data at the end

10
Text Formatting -- RTF
\rtf1\mac\ansicpg10000\uc1 \deff0\deflang1033\def
langfe1033\upr\fonttbl\f1\fnil\fcharset256\fprq
2\\panose 020b0604020202020204Arial\f3\fn
il\fcharset2\fprq2\\panose 02000500000000000000
Symbol \listoverride\listid1\listoverridecount0
\ls22\info\title Overview\author J.H.
Harrison, Jr.\operator J.H. Harrison,
Jr.\creatim\yr2000\mo7\dy11\hr21\min22\revtim\
yr2000\mo7\dy11\hr21\min22\printim\yr2000\mo6\dy
5\hr11\min2\version2 \edmins0\nofpages7\no
fwords2406\nofchars13718\\company
TUMC\nofcharsws16846\vern115\widowctrl\ftnbj
\aenddoc\formshade\viewkind1\viewscale121\viewzk2\
pgbrdrhead\pgbrdrfoot \fet0\sectd
\linex0\endnhere\sectdefaultcl \footer
\pard\plain \s16\widctlpar\tqc\tx4320\tqr\tx8640\
pvpara\phmrg\posxc\posy0\adjustright
\loch\af4\hich\af4\dbch\f4\cgrid
\field\\fldinst \cs17 \hich\af4\dbch\af4\loch\
f4 PAGE \fldrslt \cs17\lang1024
\hich\af4\dbch\af4\loch\f4 7\cs17 \par
\pard \s16\widctlpar\tqc\tx4320\tqr\tx8640\adjust
right \par \\pnseclvl1\pnucrm\pnstart1\pnind
ent720\pnhang\pntxta .\\pnseclvl2\pnucltr\pns
tart1\pnindent720\pnhang\pntxta
.\\pnseclvl3\pndec\pnstart1\pnindent720\pnhang
\pntxta .\\pnseclvl4\pnlcltr\pnstart1\pninden
t720\pnhang\pntxta ) \\pnseclvl5\pndec\pnstar
t1\pnindent720\pnhang\pntxtb (\pntxta
)\\pnseclvl6\pnlcltr\pnstart1\pnindent720\pnha
ng\pntxtb (\pntxta )\\pnseclvl7\pnlcrm\pnst
art1\pnindent720\pnhang\pntxtb (\pntxta
)\\pnseclvl8 \pnlcltr\pnstart1\pnindent720\pnh
ang\pntxtb (\pntxta )\\pnseclvl9\pnlcrm\pns
tart1\pnindent720\pnhang\pntxtb (\pntxta
)\pard\plain \qc\widctlpar\adjustright
\loch\af4\hich\af4\dbch\f4\cgrid \b
\hich\af4\dbch\af4\loch\f4 Area of Concentration
in Biomedical Informatics \par \hich\af4\dbch\af
4\loch\f4 Proposal submitted by James H.
Harrison, Jr., M.D., Ph.D. and Charles Friedman,
Ph.D. \par \hich\af4\dbch\af4\loch\f4 University
of Pittsburgh Center for Biomedical
Informatics \par \b \hich\af4\dbch\af4\loch\f4
June 16, 2000 Update\b\i \par \pard
\widctlpar\adjustright \b \par
\hich\af4\dbch\af4\loch\f4 Overview \par \fs22
\hich\af4\dbch\af4\loch\f4 Biomedical Informatics
is a rapi\hich\af4\dbch\af4\loch\f4 dly growing
subspecialty of medicine that is concerned with
biological data analysis and the management of
clinical information to optimize clinical
practice and outcomes. The field covers a broad
range of topics, from artificial intelligence and
automate \hich\af4\dbch\af4\loch\f4
d\hich\af4\dbch\af4\loch\f4 pattern
recognition, through database structure,
information encoding, management and analysis of
biomedical research information, and user
interface design, to system implementation,
outcomes analysis and clinical process
improvement. The evolution of m \hich\af4\dbch\af4
\loch\f4 e\hich\af4\dbch\af4\loch\f4 dical
practice and the ongoing dramatic improvements in
hardware and software for information processing
combine to make Biomedical Informatics one of the
most rapidly-developing and exciting areas in
Medicine. Students who have interests and/or
skills in \hich\af4\dbch\af4\loch\f4
\hich\af4\dbch\af4\loch\f4 computing and
information management may gain an introduction
to Biomedical Informatics and its application at
the University of Pittsburgh School of Medicine
by participating in a proposed Area of
Concentration in Biomedical Informatics. \par
11
Generalized Markup Languages

• Markup identifies structural elements of a
  document rather than specific formatting features
• Markup is expressed as standard text sequences
  (markup "tags")
• Formatting instructions are applied separately to
  the specified document elements
• Markup tags can be human-readable

12
Markup Expresses Metadata

• Documents naturally have content and metadata
• Metadata may help specify
• Meaning of data (e.g., standard coding)
• Arrangement of data (display)
• Correct use of the data (business rules)
• Context and relationships between data elements
• Example of display markup
• Documents contain ltemphgtmetadatalt/emphgt and
  "primary" data

Tag content
Embedded "tag"
13
Heritage of Generalized Markup Languages
Internal work at IBM
TeX, nroff, troff
SGML
Tag formats DTD format Processing rules
Frameworks
1984
Many special- purpose markup languages
XML
1998
HTML
XHTML
1990
Many special- purpose markup languages
Implementations
14
A Hypertext Markup Language Document
lthtmlgt ltheadgt lttitlegtMy Documentlt/titlegt lt/headgt lt
bodygt lth1gtThis is an example HTML
documentlt/h1gt ltolgt ltligtFirst paragraphlt/ligt ltligtSe
cond paragraphlt/ligt lt/olgt ltpgtThis is the first
paragraph. It doesn't really say very much. Now
is the time for all good men to come to the aid
of their country.lt/pgtltpgtThis is the second
paragraph. The quick brown fox jumped over the
lazy dog.lt/pgt lt/bodygt lt/htmlgt
15
HTML as a Markup Language

• HTML is a lightweight, standard, simple way to
  specify document structure (and appearance)
• HTML tags are tightly tied to visual display
• Tag function is hard-coded into HTML browsers
• Complex formatting leads to complex HTML
• Limited ability to create interactive user
  interfaces
• Available tags are specified by the HTML DTD
  (cannot create new, special-purpose tags)
• Cannot represent data structures from databases
• HTML documents are difficult to maintain in large
  numbers
• Inefficient to index and search
• HTML does not provide control over data entry
• Any text is OK

16
XML's Purpose

• HTML originally specified structural components
  of documents
• HTML has evolved to become a presentation syntax
• SGML is complex and requires complex processing
  software
• XML is a simplified version of SGML designed for
  electronic document archiving and exchange
• Allows creation of special-purpose markup
  languages
• Can represent a variety of data structures and
  semi-structured data as well as metadata
• Arbitrary tag nesting, recursion and granularity
  "coarsens" granularity gracefully
• Human-readable and machine readable
• Expected to be useful for creation of
  vertical-market data-interchange standards

17
Document Markup -- HTML

• Markup is restricted to display information
  only.
• Clinical information must be contained in free
  text.
• Available tags are specified in a rigid HTML
  "DTD"

18
Document Markup -- XML

• Markup can identify specific data elements.
• Markup is flexible to accommodate new data
  elements and additional metadata.
• Metadata can be contained within tags.
• New DTDs can be created.

19
Document StructureDocument Type Description
SP.dtd (optional, internal or remote)
lt?xml version"1.0" encoding"ISO-8859-1"?gt lt!DOCT
YPE report SYSTEM "SP.dtd"gt ltreport
type"Surgical Pathology"gt lthospital
code"PUH"gtPresbyterian University
Hospitallt/hospitalgt ltdemoggt ltpat_name
hnum"000666"gt ltfirstNamegtDonaldlt/firstNamegtltlast
NamegtDucklt/lastNamegt lt/pat_namegt ltbirthdategt01/1
8/1954lt/birthdategt ltprocedure cpt"1234"gt ltpat_
phys pnum"abcd"gt ltfirstNamegtElmerlt/firstNamegt
ltlastNamegtFuddlt/lastNamegt ltdegreegtM.D.lt/degr
eegt lt/pat_physgt ltproc_namegtUpper endoscopy of
gizzardlt/proc_namegt ltproc_dategt09/09/1999lt/proc_
dategt lt/proceduregt ltclin_hx ICD"34565"gtEpigastr
ic painlt/clin_hxgt lt/demoggt ltpathologist
pnum"7896"gt ltfirstNamegtWilelt/firstNamegt ltlastNa
megtCoyotelt/lastNamegt ltdegreegtM.D.lt/degreegt lt/path
ologistgt ltbodygt ltgross SNOMED"T-987535
O-984324"gt Received in a plastic specimen cup is
a single 1/4" dia. hard brown irregular mass with
a sharp point on one side, which is placed in
decalcification medium overnight and then
sectioned through the center and submitted in
toto. lt/grossgt ltmicro SNOMED"H-893491"gt Histologi
c appearance consistent with bone
fragment. lt/microgt ltdiagnosis SNOMED"E-8947317"
ICD"87532" CDM"reouw"gt Bone fragment consistent
with xenostosis gizzardii. lt/diagnosisgt lt/bodygt lt/
reportgt
lt?xml version"1.0" encoding"ISO-8859-1"?gt lt!ELE
MENT report (hospital, demog, pathologist,
body)gt lt!ATTLIST report type CDATA
REQUIREDgt lt!ELEMENT hospital (PCDATA)gt
lt!ATTLIST hospital code CDATA REQUIREDgt lt!ELEMENT
demog (pat_name, birthdate, procedure,
clin_hx)gt lt!ELEMENT pat_name (firstName,
middleI?, lastName)gt lt!ATTLIST pat_name hnum
CDATA REQUIREDgt lt!ELEMENT firstName
(PCDATA)gt lt!ELEMENT middleI (PCDATA)gt lt!ELEMENT
lastName (PCDATA)gt lt!ELEMENT birthdate
(PCDATA)gt lt!ELEMENT procedure (pat_phys,
proc_name, proc_date)gt lt!ATTLIST procedure cpt
CDATA REQUIREDgt lt!ELEMENT pat_phys (firstName,
lastName, degree)gt lt!ATTLIST pat_phys pnum CDATA
REQUIREDgt lt!ELEMENT degree (PCDATA)gt lt!ELEMENT
proc_name (PCDATA)gt lt!ELEMENT proc_date
(PCDATA)gt lt!ELEMENT clin_hx (PCDATA)gt lt!ATTLIST
clin_hx ICD CDATA REQUIREDgt lt!ELEMENT
pathologist (firstName, lastName,
degree)gt lt!ATTLIST pathologist pnum CDATA
REQUIREDgt lt!ELEMENT body (gross, micro,
diagnosis)gt lt!ELEMENT gross (PCDATA)gt lt!ATTLIST
gross SNOMED CDATA REQUIREDgt lt!ELEMENT micro
(PCDATA)gt lt!ATTLIST micro SNOMED CDATA
REQUIREDgt lt!ELEMENT diagnosis (PCDATA)gt lt!ATTLI
ST diagnosis SNOMED CDATA REQUIREDgt lt!ATTLIST
diagnosis ICD CDATA REQUIREDgt lt!ATTLIST
diagnosis CDM CDATA REQUIREDgt
20
XML Document Detail
Opening tag
Element name
Attribute
ltprocedure cpt"1234"gt ltpat_phys
pnum"abcd"gt ltfirstNamegtElmerlt/firstNamegt ltlas
tNamegtFuddlt/lastNamegt ltdegreegtM.D.lt/degreegt lt/p
at_physgt ltproc_namegtUpper endoscopy of
gizzardlt/proc_namegt ltproc_dategt09/09/1999lt/proc_d
ategt ltlocation name"ER"/gt lt/proceduregt
Content
Singleton tag
Closing tag
21
DTD Detail
Element Definition
Model
lt!ELEMENT procedure (pat_phys, proc_name,
proc_date)gt lt!ATTLIST procedure cpt CDATA
REQUIREDgt lt!ELEMENT pat_phys (firstName,
middleI?, lastName, degree)gt lt!ATTLIST pat_phys
pnum CDATA REQUIREDgt lt!ELEMENT degree
(PCDATA)gt lt!ELEMENT proc_name (PCDATA)gt lt!ELEMEN
T proc_date (PCDATA)gt
Attribute Definition
22
HL7Health Level 7
Organized to create standards for the exchange,
management and integration of data that supports
clinical patient care and the management,
delivery and evaluation of healthcare services

• Founded by healthcare providers in 1987
• Version 1.0 late in 1987
• Version 2.0 late in 1988
• Versions 2.1, 2.2 and 2.3 published in 1990,
  1994 and 1997 ANSI standards
• Pragmatic approach
• Work on Version 3 (XML-based) is ongoing

23
HL7 Standard

• Protocol for exchange for healthcare information
• A framework for text messages widely used in
  healthcare system interfaces
• not an application
• not a data structure or data base specification
• not an architecture for designing health
  applications
• not a specification for a message router or
  Gateway
• not a vocabulary or coding specification

24
HL7 Abstract Messages

• Text (ASCII) format specifying many message types
• Patient admission/registration/discharge/transfer,
  queries, orders, results, clinical observations,
  billing, medical records
• Transactional model for messaging, response and
  error handling
• Data fields identified by position
• Standard DOES NOT describe the byte string
  contained in the message.

25
HL7 Transactional Model
26
Admit Message (A01)
segments, fields, components subcomponents
MSH\ADT1MCMLABADTMCM198808181126SECURIT
YADTA01MSG00001P2.3ltcrgt EVNA01198808181123
ltcrgt PIDPATID12345M11JONESWILLIAMAIII
19610615MC1200 N ELM STREETGREENSBORONC27
401-1020GL(919)379-1212(919)271-3434S PATI
D123450012M10123456789987654NCltcrgt NK1JONES
BARBARAKWIFENKNEXT OF
KINltcrgt PV11I2000201201004777LEBAUERSID
NEYJ.SURADMA0ltcrgt
27
Abstract Message StructureSegments
... is 0 to 1, ... is 1 to many, ...
is 0 to many MSH Message header NTE Notes
and comments PID Patient
identification NTE Notes and comments about
the patient AL1 Allergy data PV1
Patient Visit ORC Common Order Order
Detail chosen from OBR, RXO, RQD RQ1, ODS,
ODT NTE Notes and comments about
the order OBX Observational results
NTE Notes and comments about
results BLG Billing
28
Result Message
29
HL7 v2.x is not Plug and Play

• Cost of installing an HL7 interface 2-4 weeks of
  analyst time
• Issues
• Different implicit information models
• Misunderstanding of specifications
• No vocabulary to describe conformance except by
  detailed specs
• Significant local demands on vendors

30
Goals for Version 3

• Substantially reduce interface development time
• Clarify spec for messages
• Specified information model
• Method for conformance specification
• Support modern communications infrastructures
• Reference Information Model (RIM)
• Coherent shared information model
• Includes all content of HL7 messages
• Provides consistency to messages across usage
  settings
• 120 defined classes (May '99)

31
HL7 v3 Message SpecificationHow to get from the
RIM to a specific message structure

• Message Information Model (MIM)
• Subset of the RIM contained in a specific message
• Heirarchical Message Description (HMD)
• "Recipes" for messages define data relationships
• Messages may be encoded in any of a number of
  schemes
• Encoding formats are described in Implementation
  Technology Specifications (ITS, XML is one)
• The information content of the message is
  identical regardless of the ITS

32
Advantages of XML for Message Formatting

• The syntax handles recursion and nesting
• Variably nested structures to arbitrary depth
• More flexible than segments, fields, components
  subcomponents
• Objects (including contained objects) can be
  represented
• Relational structures can be represented
• Data files are nearly self-documenting (human
  readable)
• Software tools (parsers, etc.) are generally
  available
• A developing standard in data management and
  business communication
• "Reflected Hoopla"

33
HL7 2.3 Message Format
34
HL7 v3 Message Format
35
DICOMDigital Imaging and Communications in
Medicine
An international standard supporting the
interoperability of medical imaging devices and
communication of medical imaging data

• ACR-NEMA
• American College of Radiology
• National Electrical Manufacturers Association
• Committee formed 1983
• Standard 1.0 released 1985, 2.0 released 1988
• Initial goal external interfaces for imaging
  equipment
• Version 3
• Complete, OO redesign to support robust network
  communication ISO-based layered model

36
DICOM Overview

• A set of standards linking
• Image acquisition and management devices
  (communications)
• Image/structured text archives (static files)
• Hard copy output of imaging data
• Domains
• Radiology, Nuclear Medicine, Ultrasound
• Being adopted by Cardiology, Pathology, GI
  Endoscopy
• Network Image Management (widely implemented)
• Client-server model (service class users and SC
  providers)
• Compatible with TCP/IP

37
DICOM Version 3

• Thirteen-part specification
• Images (multiple JPEG formats) and structured
  text
• Important features
• Automatic feature negotiation between devices
• Robust object model
• Extensive data dictionary
• Conformance requirement model
• File format for offline image exchange supporting
  imaging and descriptive data

38
DICOM Observation Classes

• Structured interpretation standard
• Textual information on the results of an imaging
  procedure
• Text, codes, measurements, coordinates, etc.
• Specific domain vocabularies provided by
  DICOM-SNOMED microglossaries

39
Summary

• Standard communications protocols
• TCP/IP
• File and markup formats
• ASCII, Unicode, MS Word, RTF, HTML, XML
• Healthcare data interchange and communications
• HL7, DICOM

40
References

• Van Bemmel and Musen. Medical Informatics.
  Chapter 6, pp. 81-98. Springer-Verlag 1997.
• Korpela, J. A tutorial on character code issues.
  http//www.hut.fi/jkorpela/chars.html
• Lander, R. Introduction to markup languages.
  http//pdbeam.uwaterloo.ca/rlander/XML/intro_gml.
  html
• Rishel, W. HL7 Version 3 Message Building.
  http//www.hl7.org/Library/General/V3_Message_Buil
  ding9901_WesA.pdf
• Health Level Seven, Inc. HL7 Version 3 Statement
  of Principles, 1/22/98. http//www.hl7.org/page.cf
  m?p524
• Horiil, S.C. et al. DICOM An Introduction to the
  Standard. http//www.sbis.epm.br/pepcdrom/apresent
  /pimagmed/intro.html
• NEMA. Digital Imaging and Communications in
  Medicine (DICOM) Part 1. Introduction and
  Overview. http//medical.nema.org/dicom/1998/98_01
  dr.pdf