Clinical Guideline and Protocol Automation in Medilink

1
Clinical Guideline and Protocol Automation in
Medilink

• Applying ECA Paradigm to Supporting Protocol
  Feedback Control for Diabetes Care

• Kudakwashe Dube
• Computer Science
• DIT

2
Presentation Outline

• Introduction
• Guideline Protocols
• The ECA Paradigm
• Feedback Control for Diabetes Care
• Related Work
• HyperCare
• MLMs the Arden Syntax
• Discusssion
• Closed- and Open-Loop Control Approach
• Protocol Specification, Execution and
• Management in TOPS
• Contribution to MediLinks Diabetes Domain
• Summary

3
Introduction

• Guideline Protocols
• The ECA Paradigm
• Feedback Control for diabetes Care

4
Clinical Guidelines Protocols
Guideline systematically developed statements
to assist practitioner and patient decisions
about appropriate health care for specific
clinical circumstances (The Institute of
Medicine) a method, that identifies actions,
that are to be perfomed and that specify
conditions that govern when it is appropriate to
perform them (Pattison-Gordon et al,
1996) Protocol a more detailed version of a
clinical guideline and refers to a certain class
of therapeutic interventions (Miksch, 1999)
5
Clinical Guidelines Protocols
6
ECA Paradigm
Event-Condition-Action (ECA) when an event
occurs, check the condition and if it holds,
execute the action Dittrich et. al, 1995
7
Feedback Control for Diabetes Care

• Feedback Control
• mechanism for reaching targeted goal e.g.
  achieving appropriate BP level, glycemia control
• ECA Paradigm
• ? computer-based feedback control

8
Related Work

• HyperCare
• (Cairini et. al., 1997)
• Medical Logic Modules (MLMs)
• (Hpriscak et. al., 1994)

9
HyperCare
(Cairini et. al., 1997 includes Ceri of the
active database fame)

• Aproach
• - ECA paradigm used to support clinical guideline
  compliance
• - Uses active capability in a DBMS (Chimera) to
  achieve
• inferential capabilities of production systems
• Clinical domain
• Essential Hypertension
• Guidelines used
• WHO/International Society of Hypertension
• US Joint Nat. Committee on Detection, Evaluation
  and
• Treatment of High BP
• Representation formalism
• ECA rules is used to represent protocol logic
• Uses event-based rule stratification model for to
  ensure termination
• ECA rule base created during systems analysis

10
HyperCare
Limitations

• Rule management
• manipulation
• versioning
• customisation

11
Medical Logic Modules (Mlms)

• When is the MLM pertinent?
• data storage
• another MLM
• application

• Set of medical criteria or algorithm
• concludes TRUE or FALSE

• Executed when logic concludes true
• store message
• send e-mail
• return value

MLM
Mapping MLM terms to entities within a
healthcare institution
12
Medical Logic Modules (MLMs)
Limitations

• need to interact with a clinical database to
  provide
• alerts and reminders
• curly braces problem explicit isolation of
  references to local data (standardisation
  problem, not unique to MLMs)
• no notification mechanisms for alerts and
  reminders (left to local implementation, curly
  braces problem)

13
Medical Logic Modules (MLMs)
Further Limitations

• MLM repositories are in text format
• MLMs difficult to maintain
• MLMs cannnot be queried
• automatic customisation not possible
• customisation institutional-specific,
  patient-specific not possible
• the only scope possible is global all patients,
  never individual patient in a category

14
Discussion

• Open- and Closed-Loop Control Approach
• Guideline/Protocol Automation in TOPS
• Contributions to MediLink

15
Closed vs Open Loop Systems for Diabetes care

• Closed-loop systems
• sense state of patient then deliver intervention
  without
• clinician intervention. Sensory Problems
• Open-loop systems
• clinician placed between computer-suggested
  intervention and
• the delivered patient-care task. Adopted in TOPS
• In diabetes care
• many important variables are objective ?
  open-loop control
• Possible uses of open-loop control systems
• - suggest insulin dosage adjustments
• - suggest treatment of hypercholesterolemia
• - remind clinician of various mellitus

16
Open-loop Control in the Management of Diabetes
Requirements

• exact specification of diabetes management rules
  
• substantial time commitments by diabeticians
• primary studies and decision analysis to define
  optimum use
• of interventions
• translation of existing knowledge into precise
  guidelines or
• protocols

17
Supporting Protocol/guideline Automation in TOPS

18
Protocol Specification in TOPS
Once fasting blood glucose is consistently above
6.1mM a person is deemed to have Impaired Glucose
Tolerance (IGT). These individuals do not have
Type 2 diabetes but are likely to develop it over
time.
Protocol
specification
Event availability of a new fasting blood
glucose measurement Condition fasting blood
glucose consistently above 6.1 mM Action place
patient into category of patients likely to
develop Type 2 diabetes
ltprotocolgt ltrulegt lteventgtltgtFBG result
arrivallt/gtlt/eventgt ltconditiongtltgtaverage FBG gt
6.1 mM during past X periodlt/gtlt/conditiongt ltactio
ngtlt gtplace patient in Category
Xlt/gtlt/actiongt ltrulegt lt/protocolgt
Storage
Can be shared Can be executed
19
Protocol Specification in TOPS
Protocol
Identification, security management Information
Enables customisation single set of rules
selected depending on patient condition
criteria-based Sets of ECA Rules
r3
Generic Set of ECA Rules
Present in every instance of protocol
r2
r1
Allows enforcement of high-level policy and
institutional customisation
g3
g2
Global rule set (institution-wide)
g1
20
Protocol Execution in TOPS
Protocol instance
TOPS Protocol Manager
ECA Rule Set
Time-driven ECA rule set
External event
Timer
TOPS ECA Rule Mechanism
Database ECA Rule (trigger) Mechanism
TOPS Database
Database event
DBMS (Oracle8i)
21
Protocol Management in TOPS
Static Dynamic Management
add
delete
modify
query
Protocol Specifications
Protocol Executing instances
22
Summary Possible Contributions in MediLink

• capturing knowledge of a guideline/protocol
• nature into a format that allows storage,
  execution
• and management
• Supporting compliance in diagnosis and
  management
• (including patient self-monitoring)
• Basis for integration of guidelines/protocols,
  data
• collection and analysis, and the EHCR