Methods%20for%20Computer-Aided%20Design%20and%20Execution%20of%20Clinical%20Protocols

1
Methods for Computer-Aided Design and Execution
of Clinical Protocols

• Mark A. Musen, M.D., Ph.D.
• Stanford Medical Informatics
• Stanford University

2
Research problems in medical informatics involve

• Formulation of models of clinical tasks and
  application areas
• Representation of those models in
  machine-understandable form
• Development of new algorithms that process domain
  models
• Implementation of computer programs that use
  models to automate clinically important tasks

3
Protocol-based care is everywhere

• Algorithms for mid-level practitioners
• Clinical-trial protocols
• Clinical alerts and reminders
• Clinical practice guidelines

4
Some basic beliefs

• Computer-based patient records eventually will
  become ubiquitous
• Clinical protocols canand shouldbe authored
  from the beginning as machine-interpretable
  documents
• Electronic protocol knowledge bases will allow
  computer-based patient records to enhance all
  components of patient care and clinical research

5
Work in protocol-based care

• ONCOCIN (19791988)
• Clinical trials in oncology
• Therapy Helper (19891995)
• Clinical trials for HIV infection
• EON (1989)
• Reusable components for automation of protocols
  and guidelines in a variety of domains

6
Our research addresses

• Development of computational models of
• Planning medical therapy
• Determining when therapy is applicable
• Reasoning about time-ordered data
• New approaches for acquisition, representation,
  and use of medical knowledge within computers

7
EON Components for automation of clinical
protocols

• Models of protocol concepts
• Programs to plan patient therapy in accordance
  with protocol requirements
• Programs to match patients to potentially
  applicable protocols and guidelines

8
Use of an explicit model to guide knowledge entry
Model of protocol concepts
Custom-tailored protocol-entry tool
Knowledge-base authors create protocol description
s
EON
Protocol knowledge base
Therapy- planning program
Eligibility- determination program
Clinicians receive expert advice
9
Model (ontology) of protocol concepts
10
Components of the protocol model (ontology)

• Guideline ontology
• Defines abstract structure of clinical protocols
  and guidelines
• Is independent of any medical specialty
• Medical-specialty ontology
• Defines clinical interventions, patient findings,
  and patient problems relevant in a given
  specialty
• Provides primitive concepts used to construct
  specialty-specific protocols

11
An ontology

• Provides a domain of discourse for talking about
  some application area
• Defines concepts, attributes of concepts, and
  relationships among concepts
• Defines constraints on values of attributes of
  concepts

12
Model (ontology) of protocol concepts
13
Custom-tailored protocol-entry tool
14
Details of CAF chemotherapy
15
Details of CTX prescription
16
Custom-tailored protocol-entry tool Top level
17
Specifying eligibility criteria
18
Use of an explicit model to guide knowledge entry
Model of protocol concepts
Custom-tailored protocol-entry tool
Knowledge-base authors create protocol description
s
EON
Protocol knowledge base
Therapy- planning program
Eligibility- determination program
Clinicians receive expert advice
19
Automation of protocol-based care requires

• Ability to deal with complexity of patient data
  (e.g., time dependencies, abstractions, missing
  data)
• Ability to deal with complexity of protocol
  actions (e.g., actions which are themselves
  protocols)
• A scalable and maintainable computational
  architecture

20
The EON Architecture comprises

• Problem-solving components that have
  task-specific functions (e.g., planning,
  classification)
• A central database system for queries of both
• Primitive patient data
• Temporal abstractions of patient data
• A shared knowledge base of protocols and general
  medical concepts

21
EON is middleware

• Software components designed for
• incorporation within other software systems
  (e.g., hospital information systems)
• reuse in different applications of protocol-based
  care

22
Components of the EON architecture
Therapy- planning component
RÉSUMÉtemporal- abstraction system
Chronus temporal database query system
Eligibility- determination component
Clinical information system
Patient database
Tzolkin database mediator
Protocol knowledge base
Domain model
23
Therapy-planning component

• Takes as input
• Data from computer-based patient record
• Knowledge of clinical protocol
• Generates as output
• Therapeutic interventions to make
• Laboratory tests to order
• Time for next patient visit

24
Episodic skeletal-plan refinement
1. Flesh out standard plan from skeletal plan
elements
?
2. Query database for presence of
relevant patient problems
3. Revise plan based on problems identified
25
Domain knowledge derives from knowledge base
26
Problem-solving knowledge automates specific tasks
Domain knowledge Problem-solving
method Intelligent behavior
27
Problem-solving methods

• Are reusable, domain-independent software
  components that solve abstract tasks (e.g.,
  planning, classification, constraint
  satisfaction)
• Represent data on which they operate as a method
  ontology (model), which must be mapped to the
  domain ontology that characterizes the
  application area

28
Mapping domain ontologies to problem-solving
methods
Problem-Solving Method
Method
Method
Output Ontology
Input Ontology
Domain Ontology (e.g., clinical protocols)
29
Problem-solving methods can automate a variety of
tasks

• Some skeletal planning tasks
• Therapy planning for protocol-based care (EON)
• Administration of digoxin in the presence of
  possible toxicity (Dig Advisor)
• Designing experiments in molecular genetics
  (MOLGEN)
• Each application entails mapping a different
  domain ontology to the same, reusable
  problem-solving method

30
Components of the EON architecture
Therapy- planning component
RÉSUMÉtemporal- abstraction system
Chronus temporal database query system
Eligibility- determination component
Clinical information system
Patient database
Tzolkin database mediator
Protocol knowledge base
Domain ontology
31
Our goals for eligibility determination

• Automated clinical-trial screening from
  institutional and regional databases
• Identification of specific actions that providers
  can take to enhance patient eligibility for
  guidelines and protocols
• Minimization of inappropriate enrollment of
  patients who are not eligible

32
EON eligibility-determination component (Yenta)

• Takes as input
• Computer-based patient record data
• Knowledge of eligibility criteria of applicable
  protocols
• Generates as output
• List of patients potentially eligible for given
  protocols
• List of protocols for which given patients
  potentially are eligible

33
Classification of eligibility criteria for
clinical trials

• Stable (e.g., having received prior therapy)
• Variable (e.g., routine lab data)
• Controllable (e.g., use of a given drug)
• Subjective (e.g., likelihood of compliance)
• Special (e.g., lab data requiring invasive or
  expensive tests)

34
Qualitative eligibility scores
For each eligibility criterion, for each point
in time, the computer assigns a score

• P meets the criterion
• PP probably meets the criterion
• N no assumption can be made
• FP probably fails the criterion
• F fails the criterion

35
(No Transcript)
36
Eligibility criteria derive from the electronic
knowledge base
37
Use of an explicit model to guide knowledge entry
Model of protocol concepts
Custom-tailored protocol-entry tool
Knowledge-base authors create protocol description
s
EON
Protocol knowledge base
Therapy- planning program
Eligibility- determination program
Clinicians receive expert advice
38
Components of the EON architecture
Therapy- planning component
RÉSUMÉtemporal- abstraction system
Chronus temporal database query system
Eligibility- determination component
Clinical information system
Patient database
Tzolkin database mediator
Protocol knowledge base
Domain model
39
Tzolkin database mediator

• Serves as a common conduit for all problem
  solvers that must access patient data
• Embodies components that address significant
  problems in temporal reasoning
• RÉSUMÉTemporal abstraction
• ChronusData query and manipulation

40
RÉSUMÉ temporal-abstraction method

• Takes as input primary patient data and
  previously determined abstractions of those data
• Generates as output further abstractions of the
  input
• Requires a separate knowledge base of clinical
  parameters and their properties

41
The temporal-abstraction task
42
Knowledge required for temporal abstraction

• Structural knowledge(e.g., definitional
  relationships among lab tests and clinical
  states)
• Classification knowledge (e.g., how numeric
  values map into qualitative ranges)
• Temporal-semantic knowledge(e.g., whether
  intervals are concatenable or downward
  heriditary)
• Temporal-dynamic knowledge(e.g., minimal values
  for a significant change, functions to predict
  persistence of a value over time)

43
Acquiring temporal-abstraction knowledge for
RÉSUMÉ
Model of clinical parameters
Tool for entry of temporal-abstraction knowledge
Knowledge-base authors enter knowledge required
for temporal abstraction
TZOLKIN
Parameter knowledge base
RÉSUMÉ temporal-abstraction system
Abstractions of relevant clinical parameters
44
The EON Architecture

• Problem-solving components that have
  task-specific functions
• A central database system for queries of both
• Primitive patient data
• Temporal abstractions of patient data
• A shared knowledge base of protocols and general
  medical concepts

45
A protocol model shared among all components

• Makes explicit relevant assumptions about the
  application domainwe know what our programs know
• Consolidates the task of maintaining the domain
  knowledgeall the knowledge is in one place and
  can be examined in a coherent fashion

46
Planned applications of EON

• Hypertension guidelines at Palo Alto VA Health
  Care System
• Fast Track Systems, Inc., plans to develop
  systems for automation of clinical trials

47
EONs component-based approach allows

• Developers to create new problem-solving modules
  that plug and play
• Clinicians to create new guideline knowledge
  bases that can interoperate immediately with
  existing components
• System architects to integrate components with
  other software modules using standard
  communication methods

48
Some implications of our work

• Enhanced authoring, maintenance, and execution of
  clinical protocols and guidelines
• Incorporation of guideline-based practice into
  routine patient care
• Increased participation of community-based
  practitioners in clinical research