Post on 30-Dec-2015
description
PROcesses &
Editable byUSers Hemant Shah M.D., M. Surg.
Scientist, Information Sciences
City of Hope National Medical Center
Duarte, CA
hemant@proteme.org
Transactions
A Guidelines Model
In This Presentation …
A very brief introduction of Proteus
Demonstration of software tool Protean and other associated tools
Implications for healthcare
Potential uses of Proteus technology in healthcare tools
Proteus – A Brief Introduction
What Is Proteus?
A model for constructing clinical decision-support Guidelines with entities called Knowledge Components which are:ExecutableEditableReusable
Proteus is a Model for…
Creating clinical practice guideline based decision-support systems
EMR systems
Kernel of integrated healthcare information systems
Proteus Contains
A specification of an architecture forExecutable GuidelinesSystems to handle them
A notation system for Guidelines – human & machine readable
The Vision
The Vision
Organization B
Knowledge Component Server
The Vision
Knowledge Component
Server
Organization A
What is a knowledge component (KC)?
A software component with a discrete bit of knowledge Complete in itself Can manage its own internal affairs Can be “connected” with other KCs to work
cooperatively with them
Contains knowledge about a clinical activity: Actions to be performed Events to look for Data to be collected from the actions and events Interpretation and implications of that data Supplementary information about the activities
(e.g. links to websites)
Knowledge Component
KC Represents Clinical Process
(e.g. diagnosis of acute abdomen pain)
Clinical Transaction
Knowledge Component (KC)
Transaction may be Clinical Event
(e.g. vomiting) Clinical Action
(e.g. Palpation of liver)
KC may contain data-fields describing the underlying
clinical entity
Lump
Tenderness
Vomiting
Temperature
Abstraction
Value of KC
Knowledge Component
KCs can be NestedTo represent composite processesTo reduce complexity
KC to Guidelines
KCs can be linked by Activity-links To represent process To define Guidelines
Guidelines to EMR
Lump presentTenderness severeVomiting yesTemperature 102 F
Instantiated (executed) KCs become medical record
Knowledge Component
Lump
Tenderness
Vomiting
Temperature
Abstraction
Inference toolPart of KC, yet separate
Just an interface Technology neutral Pluggable
Decides Abstraction – The value of
the component Activity within the
component
Pluggable Inference Tool
Test ATest A
Test BTest B
Test CTest C
Action AAction A
Action BAction B
User’s System
Knowledge Component
Inference ToolInference Tool
NetworkInference Tool
Inference Tool
Internet
Inference Tools•Algorithm•Decision Tables•Decision Theory•Rule Based System•Neural Network•Fuzzy System•Patient assisted decisions•Human expert (even user)•User Defined•User Specified •Combination of these
Inference Tools•Algorithm•Decision Tables•Decision Theory•Rule Based System•Neural Network•Fuzzy System•Patient assisted decisions•Human expert (even user)•User Defined•User Specified •Combination of these
Inference tool reference
Two Types of Knowledge Components
Transaction KC
Process KC
Transaction KC
Represents: Action or Event or combination
Contains: Data Elements Abstraction
Inference tool
Transaction NameTransaction Value
Transaction NameTransaction Value
Data 1 - Value
Transaction Icon
Data 2 - Value Data n - Value
Transaction KC Icon
Process KC
Represents Clinical process
Contains Nested KCs Activity Links Abstraction
inference tool Action inference
tool
Process NameProcess Value
Process NameProcess Value
Process Icon
ProcessProcess
TransactionTransaction
TransactionTransaction
Nested KCs
Links
Process KC Icon
Activity Links
Represent the sequence of Triggering of KCs and how they are triggered
Inferential Link
Sequential Link
Synchronous Link
Inferential Stop Link
Sequential Stop Link
Activity Links
Proteus Model – UML Class Diagram
Proteus Model – UML Class Diagram – Proteus
Guideline Component Class
11
2
Execution and Inference
2
AA BB
33
DD EE
CC
A B
The Cycle is repeated
Process KC 2’s inference tool decides next action.
Guideline’s abstraction is changed
Process KC 1’s abstraction is changed
Process KC 2’s abstraction is changed
Transaction KC A is executed
D E
Process KC 1 is executed
Process KC 2 is executed
Protean – A Software Environment for Proteus Guidelines
And Other Ancillary Tools
Tutorial
To test some of the Proteus concepts in action, see the tutorial
http://www.proteme.org/tutintro.html
Features of Protean
Loading and Display of Guideline
Execution Inferencing and Decision Support
What actions to performWhat events to look for Interpretation based on the dataSupplementary information
Data Entry SupportEMR
Features of Protean
EditingCreating New ElementsDeleting Modifying existing elementsReuseChanging the Inference tool Changing the inference tool behaviorUMLS Knowledge Source Server access to
associate an entity with a UMLS term
Extensibility – JIT feature
A patient has been selected as shown by the title of the Protean window.Here the user in the process of selection of a guideline for the patient.
The guideline is loaded in Protean.
Guideline has been “started”. The first Process KC within it, “Magsulf Loading” gets triggered first, which leads to the triggering of the first Transaction KC, “Convulsions eval” in it. The Transaction KC is shown as a dialog box for the user to enter the data in it.
The guideline is shown partly executed here. Since the activity link between “Intravenous” and “Intramuscular” is a sequential one, anytime “Intravenous” is executed, after it terminates, the “Intramuscular” Transaction KC is always also triggered. Here a wait of twenty seconds has been specified for the activity link. Which means only after the duration has elapsed will the next KC be triggered.
The yellow box is countdown clock telling the user the time remaining before the next activity is triggered. The edge that connects the box to the link, tells the user where the execution is stalled.
Here an incomplete guideline is loaded in Protean. The user is in the process of completing it. The tree on the left shows all the KCs that are available in the repository . The user drags a Process KC, “Diabetes Diagnosis” from the repository on the guideline.
The user has dragged the Process KC, “Diabetes Diagnosis” and has dropped it on the guideline.
The user creates a link between the pre-existing Process KC, “Management of PROM” and the newly dropped Process KC by dragging from one to the other. A dialog box opens up to allow the user to specify characteristics of the behavior of the new activity link.
The new link has been created
This shows the rule editing application GREEd (Graphical Rule Elements Editor). The Process KC, “Diabetes Diagnosis” is loaded into the application. The list on the top left shows all the rules in the KC. Using simple drag and drop actions the user has created a rule here.
The main panel shows all the KCs contained in the “Diabetes Diagnosis” Process KC, on the left side. The right side of the main panel shows all the values (abstractions) that the “Diabetes Diagnosis” KC can possessThe rule is shown as Java code in the lower panel. Since the interpreter for this rule is BeanShell which interprets Java as a script – the Tab says BSH rule.The User-view tab shows the same rule in user readable format.Tabs for Arden Syntax and Jess are being constructed
This shows the “Magsulf Loading” Process KC being edited. The user can type a broader term in the Term field, and click the [><] button to connect to UMLS Knowledge Source Server over the Internet to select a more specific term. This feature allows every KC to be tagged with a concept in an Ontology/Vocabulary. The advantage is to index and search for KCs in a repository and to allow features like Just in Time information retrieval.
Extensibility for Non-Clinical Functionality
Data
Action
Data DecisionActionDecision
ActionCore (clinical) Process
Action
Associated Process
Clinical Process as a Skeleton
Almost everything in healthcare can be mapped to the elements of the Clinical Process
Proves that clinical process is the core
Gives unlimited extensibility
Layers for Unlimited Extensibility
Physician
Researcher
Administrator
Accountant
Organization (a)
Expert
Knowledge Component
Servers
IndependentClinician
Organization (b)
System Overview
“Publish” KCs
Access KCs
Human Expert as an “inference tool”
Inference tool (b)
Inference tool (a)
EMR
Get KCreferences
Access
Knowledge Managers
InferenceTools
Naming Server
Knowledge Users
Healthcare Delivery Organization
Comparison with Other Approaches
Other models – Modularity Missing
Inaccurate lines of separation
•Workflow entities are part of the clinical activity entities.•Inferencing elements are part of clinical activity elements
PROforma PROforma
Attributes of the generic task
Attribute Description
Name Unique identifier of task
Caption Descriptive title of task
Description Textual description of task
Goal Purpose of task
Pre-conditions Conditions necessary before a task may be started
Trigger conditions Conditions which will initiate a task
Post-conditions Conditions true on task completion
Other models – why we cant have modularity
Inaccurate lines of separation
•Programmatic (inferencing) entities are represented as steps just like clinical activity entities.
GLIFGLIF
Other models – why we cant have modularity
Inaccurate lines of
separation•next_step is an attribute in each step which is directive for the next action.
GLIFGLIF
Other models – why we cant have modularity
Inaccurate lines of separation
•Sequential Step has attribute followed_by within the Guideline entities.
EonEon
Object oriented design principles
Encapsulation and information hiding
Abstraction
Design - implications
Proteus Others
Independent development of entities Different inference technologies
Pluggable inference tools Ease of editability
How systems based on Proteus can change the way healthcare is conducted?
Implications
Bridging the Research to Practice GapIn the beginning, there was – The Gap
ResearcResearchh
The Gap
PractisePractise
Bridging the Research to Practice Gap
Bridging the Gap – The Last Mile Problem
ResearcResearchh
The Gap narrowed
Access Access to to
ResultsResultsPractisePractise
Bridging the Research to Practice Gap
Bridging the Gap – The Last Mile Problem
ResearcResearchh
The Gap still exists
Meta-Meta-ResearcResearc
hh
Access Access to to
ResultsResultsPractisePractise
Bridging the Research to Practice Gap
Bridging the Gap – The Last Mile Problem
ResearcResearchh
The Gap
Meta-Meta-ResearcResearc
hh
Access Access to to
ResultsResultsProteusProteus PractisePractise
Implications Clinicians
Decision-support for individual patients at point of careKnowledge representation that clinician understands and can manipulate and that is not passive but executableFacilitating visualizing of the previous data which has a bearing on the current decision to be madeUsage of contexts for the decision on handVisualization of the potential courses for a patientOffering suggestions for next action to be takenDirect application of principles of Evidence Based Medicine (EBM)Link to reference medical literature pertinent to action or decision being takenUse of knowledge created by other experts or institutesAccess to real-time expertise of other experts in a collaborative manner, even if the experts are spread over different locationsAllowing invoking appropriate Artificial Intelligence toolsAllowing multiple inferencing technologies to be used
Decision supportDecision supportDecision supportDecision support
Implications Clinicians
Data-entry supportAbility to manage different conditions, even new onesAbility to use data of colleaguesAllow usage of physician’s data by othersPermit scientific audit for performance review or for medico-legal purposesAbility to create their own protocols – increases participation in guidelines and in sustaining the systemAccess to patient information in several ways, including webSafeguard against litigations by helping make proper decisions automatic logging of things like
Actions Times Persons Deviations from the standard action for the situation
EMR advantagesEMR advantages EMR advantagesEMR advantages
Implications Clinicians
Ability to deal with uncertainty and unexpected situations so common in clinical practice
Protection against usage of outdated concepts by allowing usage of protocol components that are updated by other experts or organisations, without any effort on part of the user
New way of professional remuneration that also takes into consideration intellectual content of their clinical activities rather than just their qualifications, experience or the procedures performed
Ability to use many software concepts and tools of Medical Informatics besides those for decision-support seamlessly, like electronic medical record (EMR), controlled vocabulary sources, telemedicine tools etc.
Other advantagesOther advantages Other advantagesOther advantages
Implications for ManagersStandardized healthcare, QAHealthcare quality evaluationAccurate way of estimating, monitoring and predicting Costs, Requirements, Reimbursements, Revenues, patient satisfaction etc.Executive decision-support systemsDisease management, case management, outcomes management, integrated healthcare delivery frameworkCost containmentContinuity of healthcare
Implications for PatientsPatient-education regarding her condition and specific features of her own case
Informed consent
Allow patient aided management
Patient participation in management of her condition
Implications for Publishers
A new way of publishing medical literature linked directly with medical management
Not passive knowledge but dynamic executable type
Ability to assess “usage” of information as compared to “access” of information
For LibrariansProvide accurate information relevant to the cases being managed
Linking it to the appropriate steps in the management
Anticipate what information will be required by the clinicians and providing it before that step in management is reached
Implications for
Researchers Research utilization directly by clinicians directly plugging in their results or discoveries into clinical protocols
Compliance with research protocols, while allowing clinicians managing the cases to change parts which do not compromise the research
Ability to study diseases and their managements in a different way – as processes
Implications for MI & IT FolksDevelopment of new applications and knowledge components for healthcare
Extensibility in applications and in protocols including non-clinical extensions
Standard components based on CORBA, which will have interoperability with other components
Potential Tools
Proteus Based Tools for Clinical Trials
Eligibility determinationProtocol prioritizationDose Escalation & Maximum Tolerable Dose determinationAdverse Event Determination Grading Monitoring
Post-approval research behavior change monitoring
CommunicationEducation
Proteus Based Tools for Clinical Medicine
Guideline based Clinical Decision Support System with a library of guidelines for set of conditions
Electronic Medical Record
Proteus Based Tools for Clinical Trials
Clinical Trials Workflow Designer and Executor
Proteus Based Tools for Education
Guidelines based teaching of clinical care for specific conditions
Conclusion
Something like Proteus is imperativeSomething like Proteus is imperative