An Event-Driven Approach to Computerizing Clinical Guidelines Using XML

1
An Event-Driven Approach to Computerizing
Clinical Guidelines Using XML
Presented By Essam Mansour School of
Computing Dublin Institute of Technology Ireland
2
Presentation Outline

• Introduction
• Computerised Clinical Guidelines (CGs)
• Related Work
• Shortcomings
• Project Aim and Objective
• Event-Driven Approach and Framework for
  Computerising CGs
• AIM-SL An Event-Driven Guideline Specification
  Language
• An Event-Driven Guideline Execution Model for the
  Patient Plan
• AIMS A Prototype System for Managing CGs
• Future Work
• Summary

3
Introduction
Computerised Clinical Guidelines

• Clinical Guidelines (CGs) (Field 1992) provide
  guides for doctors and patients in determining
  recommended strategies for monitoring the
  patients condition.
• Clinical Guidelines as Active Information
• The management aspects required for CG automation
  are
• Specification
• Customization
• Execution
• Query and Manipulation
• Distributed Collaboration and Sharing

4
Introduction
Related Work

• Computerised Clinical Guidelines Approaches
• Mainly, focus on specifying and executing
  CGs(Clercq 2004)
• They adopt several techniques, such as AI and DB
  
• Clinical Events
• Detect and respond to clinical events (McDonald
  1976)
• ECA Rule paradigm
• Arden Syntax (Clayton 1989)
• HyperCare (Caironi 1997)
• XML
• XML-based languages for formalising CGs (Dubey
  2000, Fox 2001, Greenes 2001)
• Facilitates the sharability and distribution
• Incorporating ECA rule into XML
• Active XQuery (Bonifati 2002)
• An Event-Condition-Action language for XML
  (Bailey 2002)
• AXML (Abiteboul 2002)
• XChange (Bry 2005)

5
Introduction
Shortcomings

• In existing CG approaches, there are little or
  no
• Support for Clinical events.
• Support for manipulating and querying clinical
  guidelines information.
• Temporal support for the computerized clinical
  guidelines.
• Support for sharing the execution history of the
  CGs.
• The event-driven approaches to computerising CGs
• Arden Syntax does not distinguish between the
  generic CG and its generated instances
• Arden Syntax Specifications are stored as text
  file.
• HyperCare does not provide a generic mechanism to
  be applied to other CGs.
• In providing an implemented triggering mechanism
  for a native XML DBMS.
• In supporting temporal triggers over XML
  documents.

6
Introduction
Project Aim and Objectives

• The project aims at developing an information
  management framework and approach to
  computerizing clinical guidelines.
• The developed framework and approach should
• provide support for specifying, executing,
  manipulating and querying clinical guidelines
• Facilitate the distribution of CGs among
  heterogeneous systems
• provide temporal support for the computerized
  clinical guidelines
• Facilitate the integration of the clinical
  guidelines system into the electronic healthcare
  record (EHR) system.

7
Event-Driven Approach and Framework
Our event-driven approach to computerising CG
Facilitates the distribution and sharing
XML-Based Clinical Events
Clinical Events
XML
Event Driven Approach for CG Computerisation Using
XML
XML-Based Clinical Guidelines
Clinical Event supported by ECA
Clinical Guidelines
ECA Rule Paradigm
Event Driven Based for specifying and
executing Clinical Guidelines
Facilitates the integration into EHR
8
Event-Driven Approach and Framework
Our event-driven approach

• Relies on the ECA rule paradigm and XML
  technology.
• Provides the infrastructure for an
  Internet-enabled unified framework for CG
  management
• Event-driven Specification Language for CG
• ECA rule execution mechanism
• Easy integration into EHR management system
• Manipulation and Query language
• Easy dissemination

9
Event-Driven Approach and Framework
The SpEM Framework
Query, Manipulation Customization
Specification Plane
Manipulation Plane
Instantiation Change Propagation
Execution Plane
Query, Manipulation Execution Replay
10
An Event-Driven Specification Language
Active Information Management (AIM) Language for
Clinical Guidelines
High-level Declarative language
AIM
AIM-SL
AIM-QL

• - Supports Specification Plane
• Model
• Formal language
• Is based on ECA rule and XML

• - Supports The Manipulation Plan
• Customization
• Query and Manipulation
• Execution Replay
• Extends the XQuery language

11
An Event-Driven Specification Language
The XML Schema Model for AIM-SL
Element
Attribute
Sequence
12
An Event-Driven Specification Language
AIM-SL Specification for Microalbuminuria
Screening (MAS) Protocol

• -ltprotocol id"ProID-MAS"gt
• ltnamegt microalbuminuria screening (MAS)
  protocol lt/namegt
• ltcategoryIDgtCID316lt/categoryIDgt
• ltheadergt
• -ltSchedulesgt
• -ltschedule id"SIDMAS"gt
• ltnamegtBasic MASlt/namegt
• ltheadergt
• -ltscheduleRulesgt
• ltrule id"MAS1"gt
• ltrule id"MAS2"gt
• lt/scheduleRulesgt
• lt/schedulegt
• lt/Schedulesgt
• lt/protocolgt

13
An Event-Driven Specification Language
AIM-SL Specification for MAS Protocol
-ltrule id"MAS2"gt ltnamegtRule 2 of basic
MASlt/namegt ltpropertiesgt ltheadergt
- ltbodygt ltOntologyMapping gt
-ltevent id"E2.1"gt ltongt
ltepisode id"E2.1"gtTestResultRecievedlt/episodegt
lt/ongt lt/eventgt ltcondition
id"ID36"gt - ltaction id"AID36"gt
- ltdogt -ltAIM-QLActiongt -
ltaddgt ltrule id"RID3"gt
ltrule id"RID4"gt lt/addgt
lt/AIM-QLActiongt lt/dogt
lt/actiongt lt/bodygt lt/rulegt
Rule MAS2 ON receiving the result of test ACR
IF the ACR result is greater than 25
DO Add rule 3 and 4 to order a ACR test at
day number 6 and 38
Customization Process
lt/OntologyMappinggt ltterm id"E2.1"gt lteventgt
ltnamegtTestResultRecievedlt/namegt
ltDBOperationgtinsertlt/DBOperationgt
ltdocumentNamegtAIMS_TESTORDERRESULTlt/documentNamegt
ltnodePathgt/TestResult testNameACRlt/nodePa
thgt ltnodeNamegtTestResultlt/nodeNamegt
lt/eventgt lt/termgt lt/OntologyMappinggt
14
An Event-Driven Execution Model
Patient Plan

• Is an instance of a protocol for a specific
  patient.
• Has an active part and a passive part
• Is dynamically changing
• grows over time

-ltpatientPlan forPatientNo"PIDA" generatedFrom"
ProID-MAS " startTime"1" endTime"Now"gt
-ltschedules startTime"1" endTime"Now"gt -
ltschedule schIDREF"sch1" startTime"1"
endTime"Now"gt -ltscheduleRulesgt
ltrule rulIDREF"MAS1" startTime"1"
endTime"Now"gt lt XQueryTrigger gt
-ltrule rulIDREF"MAS2" startTime"1"
endTime"Now"gt ltXQueryTriggergt
lt/scheduleRulesgt lt/schedulegt
lt/schedulesgt lt/patientPlangt
The Passive Part That logs the history of the
execution
The Active Part That shall be registered in the
system
15
An Event-Driven Execution Model
Event Manager
Register an event
Patient Plan
Notify
Related Time Event
Absolute Time Event
Episode Event
Active Part
Passive Part
Event
Condition
Action
Log the changes
External
Internal
Notify
Notify
Register an event
Register an event
CG Action
HM Action
External DB
Internal DB
16
An Event-Driven Execution Model
Rule MAS2 ON receiving the ACR test result
IF the ACR result is greater than 25
DO Add rule 3 and 4 to order ACR test at day
number 6 and 38
Rule MAS1 ON day 2 of the patient admission,
DO order the ACR test.
Day 2
What is PP in Day 4?
Day 3
1-Now
1-Now
1-2
1-3
Replay PP from Day 3 to 5
1-3
1-2
1-Now
1-Now
1-Now
1-Now
Temporal Query Support
1-2
1-3
17
AIMS A Prototype System for Managing CPGs
External Systems and Users
This layer allows user to manage the CPGS and the
external system to supply and receive
information.
Clinical Guidelines Management Module
DB Access Module
The ECA Rule Extension Module
This layer extends the Active DBMS to support
time triggers, and external communication
mechanism
Time Trigger Manager
Listener
Message Processor
AIMS Manager An Active DBMS
An Active DBMS, such as Oracle and DB2, provides
support to store and retrieve XML
Notifier
Active EHR
The XML Repository
ECA Rule Mechanism
Patient Plan
EHR
AIM-SL Specification
The XML Repository could be queried using
SQL/XML or XQuery
SQL/XML Triggers
SQL/XML Procedures
18
A Prototype System for Managing CPGs
Query and Manipulation
-ltrule id"MAS2"gt ltnamegtRule 2 of basic
MASlt/namegt ltpropertiesgt ltheadergt
- ltbodygt ltOntologyMapping gt
-ltevent id"E2.1"gt ltongt
ltepisode id"E2.1"gtTestResultRecievedlt/episodegt
lt/ongt lt/eventgt ltcondition
id"ID36"gt - ltaction id"AID36"gt
- ltdogt -ltAIM-QLActiongt -
ltaddgt ltrule id"RID3"gt
ltrule id"RID4"gt ltrule
id"RID5"gt lt/addgt
lt/AIM-QLActiongt lt/dogt
lt/actiongt lt/bodygt lt/rulegt
Update Rule 2 by adding rule 5 to its action.
Oracle SQL/XML Update statement
UPDATE aims_protocol_tab SET OBJECT_VALUE
insertChildXML(OBJECT_VALUE, '//scheduleRules/ru
le2//AIM-QLAction/add', 'rule',
XMLType(Rule5)) WHERE existsNode(OBJECT_VALUE, '//
scheduleRules/rule2//AIMQLAction/add') 1
We might need to propagate this changes to the
generated patient plan.
19
A Prototype System for Managing CPGs
Query and Manipulation
Select p.OBJECT_VALUE.extract('/patientPlan_at_PatNo
''PIDA''//rulestate/value''ready''') From
aims_pplan_tab p
Retrieve all the schedule rules, which are
currently ready for execution
1-2
1-3
X
3-Now
3-Now
MAS1
MAS2
MAS4
MAS3
1-2
1-3
3-Now
3-Now
state
state
state
state
1-2
3-Now
3-Now
1-3
Value
Value
Value
Value
ready
ready
ready
ready
20
A Prototype System for Managing CPGs
Query and Manipulation

• The modern active DBMS, such as Oracle, provides
  the primitive support to query and manipulate
  the CGs information.
• A high level query and manipulation language is
  required
• Temporal query support is needed

21
Future Work

• AIMS A proof-of-concept system.
• Developing AIM-QL
• Extending XQuery to provide a temporal query
  support
• Distributed Management for the patient plan
• Evaluating AIMS system

22
Summary

• A background, related work, and shortcoming
• The project aim and objectives
• Our event-driven approach and framework for
  computerizing CG.
• AIM-SL A specification language for CGs.
• The execution model for the patient plan
• AIMS A Prototype System for Managing CGs

23
T
u
o
h
a
k
n
Y
24
An Event-Driven Approach to Computerizing
Clinical Guidelines Using XML
Authors Essam Mansour Bing Wu Kudakwashe
Dube Jian Xing Li