1 |
P a g e
GUIDELINE REPRESENTATION
AND EXECUTION TOOLS
An Evaluation Study
MAGDALENA GAMBA
MASTER THESIS
SUPERVISORS:
Stephanie ‘Ace’ Medlock, DVM, PhD
Danielle Sent, PhD
UNIVERSITY OF AMSTERDAM
2 |
P a g e
Student:
Magdalena Gamba
Student Number: 10916520
Email:
gamba_m@yahoo.com
SRP Mentor:
Stephanie “Ace” Medlock, DVM, PhD
Department of Medical Informatics, AMC
University of Amsterdam
SRP Tutor:
Danielle Sent, PhD
Department of Medical Informatics, AMC
University of Amsterdam
SRP address:
Department of Medical Informatics, AMC
University of Amsterdam
Meibergdreef 15
1105 AZ Amsterdam
Period:
3 |
P a g e
Table of Contents
Abstract ... 4
1. General Introduction ... 5
2. Literature Review ... 8
Introduction ... 8
Methods ... 9
Results ... 10
Arden Syntax ... 12
PROforma ... 13
Prestige ... 13
EON ... 14
PRODIGY ... 14
GASTON ... 14
GLARE ... 15
Asbru ... 15
GLIF ... 16
GUIDE/NEWGUIDE ... 16
GEM ... 17
DeGeL ... 17
Stepper ... 18
SAGE ... 18
HELEN ... 18
Semantic Web Technologies and Inference Engines ... 18
Other ... 19
Discussion ... 23
3. Evaluation Criteria ... 27
Introduction ... 27
Methods ... 27
Results ... 28
Discussion ... 30
4. Evaluation of Tools ... 32
Introduction ... 32
Methods ... 33
Results ... 34
Discussion ... 36
5. General Discussion and Conclusion ... 38
References ... 41
Appendix A ... 55
Appendix B ... 57
Appendix C ... 70
Appendix D ... 83
Appendix E ... 86
4 |
P a g e
ABSTRACT
Background
: Various guideline representation and execution tools and decision support
systems have been developed and evaluated over the past 30 years. However, there is no current
overview available of these tools, and a comprehensive framework of evaluation criteria is
lacking.
Methods
: We conducted a broad literature review to identify currently available guideline
representation and execution tools, as well as criteria that could be used to evaluate them. We
subsequently evaluated two freely available systems in their function to represent and execute
a section of the Atrial Fibrillation guideline.
Results
: We identified 19 toolchains, of which each comprised of guideline modeling and
execution tools related to a specific representation formalism. Two out of the 19 toolchains
contained free and open source tools. 72 evaluation criteria, grouped into 17 themes, were also
identified. We evaluated Arden2ByteCode from the Arden Syntax toolchain, and DELT/A,
AsbruView and Asbru Interpreter from the Asbru toolchain. Arden2byteCode, a free and open
source tool, presented an intuitive user interface and adequate user support. Arden Syntax was
simple to learn. The tool did not provide facilities for guideline visualization or browsing, and
integration with an electronic patient record could not be simulated. The Asbru toolchain was
more complete, as it offered tools for guideline modeling (DELT/A), for visualizing guideline
logic (AsbruView) and for guideline execution (Asbru interpreter). User interfaces were basic,
user support was limited, and not all functions worked correctly. Integration with a patient
record could be simulated at a basic level. Asbru was difficult to learn due to its unique syntax
and semantics.
Conclusion
: Despite the number of guideline representation and execution tools that have
been developed, only a few are freely available, and even fewer are available as free and open
source tools. From our initial evaluation, we found Asbru toolchain to be more complete, and
Arden2ByteCode more user friendly. Our results imply that Arden2ByteCode is more suitable
for use in an educational setting, and its open source license makes it potentially more pertinent
for use in research. Further evaluation using more complex guidelines, as well as integration
within a clinical information system is recommended. While we defined and used general
criteria themes in our study, the development of a comprehensive framework would ensure a
complete and standardized evaluation process.
5 |
P a g e
ABSTRACT
Achtergrond: In de afgelopen 30 jaar zijn verschillende richtlijnenrepresentatie en
-executietools alsook beslissingsondersteunende systemen ontwikkeld en geëvalueerd. Echter,
er is geen actueel overzicht beschikbaar van deze tools en er ontbreekt een raamwerk van
evaluatiecriteria.
Methoden: We hebben een brede literatuurstudie uitgevoerd om de huidige beschikbare
richtlijnenrepresentatie en -executietools te identificeren, en ook de criteria die kunnen gebruikt
worden om ze te evalueren. Vervolgens hebben we twee vrij verkrijgbare systemen geëvalueerd
met betrekking tot de mogelijkehid om een deel van de Atrium Fibrillatie richtlijn te
representeren en uit te voeren.
Resultaten: We hebben 19 combinaties van tools geïdentificeerd die bestonden uit
richtlijnmodellering en -uitvoeringstools gerelateerd aan een specifiek representatieformalisme.
Twee van de 19 zogenaamde combinaties van tools bevatten vrije en opensource tools. 72
evaluatiecriteria, gegroepeerd in 17 thema’s, zijn geïdentificeerd. We hebben Arden2ByteCode
van de Arden Syntax toolchain, en DELT/A, AsbruView en Asbru Interpreter van de Asbru
toolchain geëvalueerd. Arden2ByteCode, een vrij en opensource tool, heeft een intuïtieve
gebruikersinterface en voldoende gebruikersondersteuning. Arden Syntax was eenvoudig om
te leren. De tool bood echter geen functionaliteiten om richtlijnen te visualiseren of erdoor te
bladeren, en integratie met een elektronische patiëntendossier kon niet worden gesimuleerd. De
Asbru toolchain was completer aangezien het tools aanbood voor richtlijnenmodellering
(DELT/A), voor het visualiseren van richtlijnenlogica (AsbruView) en voor het uitvoeren van
richtlijnen
(Asbru
Interpreter).
Gebruikersinterfaces
waren
eenvoudig,
gebruikersondersteuning was beperkt en niet alle functionaliteiten werkten correct. Integratie
met een patiëntendosssier kon op een basisniveau worden gesimuleerd. Asbru was moeilijk om
te leren omwille van zijn unieke syntax en semantiek.
Conclusie: Ondanks het grote aantal richtlijnenrepresentatie en -uitvoeringstools dat
ontwikkeld is, zijn er slechts een paar tools vrij beschikbaar en nog veel minder zijn beschikbaar
als vrije en opensource tools. Van onze initiële evaluatie vonden we de Asbru toolchain
completer, en Arden2ByteCode gebruiksvriendelijker. Onze resultaten impliceren dat
Arden2ByteCode geschikter is voor gebruik in een educatieve omgeving en zijn open source
licentie maakt het potentieel evident beter voor gebruik in onderzoek. Verdere evaluatie,
gebruik makend van complexere richtlijnen, en integratie in een klinisch informatiesysteem
wordt aangeraden. Hoewel we algemene thema’s van critera gedefinieerd en gebruikt hebben
in onze studie, zou de ontwikkeling van een uitgebreid raamwerk een compleet en
gestandaardiseerd evaluatieproces garanderen.
6 |
P a g e
1
GENERAL INTRODUCTION
Clinical guidelines are recommendations pertaining to the evaluation, diagnosis, and treatment
of specific medical conditions. They are developed by panels of clinical experts for use in
medical practice, and are based on best evidence obtained from research, as well as literature
and/or consensus. In a 2011 consensus report issued by IoM
1, guidelines are defined as
“statements that include recommendations intended to optimize patient care that are informed
by a systematic review of evidence and an assessment of the benefits and harms of alternative
care options” [1]. The report goes on to state that in order to optimize patient care, clinical
guidelines need to be trustworthy, having the potential “to reduce inappropriate practice
variation, enhance translation of research into practice, and improve healthcare quality and
safety”.
Clearing houses such as the US Agency for Healthcare Research and Quality Guideline
Clearing House (www.guideline.gov) systematically categorize guidelines into domains,
thereby helping users to identify and choose guidelines pertinent to their field of study, research
or practice [2]. Guidelines can be freely downloaded as text based documents from such
websites, however the provision of guidelines in this format is not conducive to their actual
implementation in practice. In the Netherlands, the Dutch Guideline Database (www.
richtlijnendatabase.nl), a project of the Comprehensive Cancer Centre the Netherlands (IKNL)
and the Knowledge Institute of Medical Specialists (KiMS), comprises of guidelines used
mainly in hospital settings. In addition to text based format, the database presents guidelines as
modules, where each module refers to a different aspect of a guideline. This structure makes it
easier for users to refine their searches, especially through use of keywords and filters.
Guidelines are usually composed of large documents that are cumbersome and time consuming
to read, and are difficult to update when new scientific evidence is made available [3]. Medical
practitioners may not have enough time or skill to find all the information they need or to keep
up with updates, and may not be able to recognize when information is untrustworthy or
reconcile conflicting information [4]. Terms within guidelines may not be clearly defined and
recommendations vaguely worded [5]. They may contain inconsistencies, ambiguities, and
1
Institute of Medicine (IoM), currently known as the National Academy of Medicine (NAM), is an American
non-profit non-governmental organization that provides advice on national issues concerning biomedical
science, medicine and health.
7 |
P a g e
logical errors which make the implementation of these guidelines into practice difficult [6].
Text based guidelines also contain broad information on diagnoses, diseases and treatments,
which limits their usability at point of care as they are not customized for use with specific
patients [7].
To address the issues above and facilitate the implementation of clinical guidelines into medical
practice, many representation and execution tools have been developed. Computerizing
guidelines eliminates the need for medical practitioners to read text based guidelines in their
entirety, as well as reduces the time needed for the transfer of new medical knowledge into
practice [8]. By integrating them with electronic medical records, patient tailored
recommendations can be given at point of care, increasing the probability of having an impact
on clinician behavior as opposed to using only narrative guidelines [9]. Additionally, the
process helps in the clarification and verification of consistency, logical structure, and
completeness of clinical guidelines [6].
Despite the clear benefits of computerizing guidelines, there are challenges that occur.
Transforming text based guidelines into computerized representations requires medical
knowledge to be manually extracted, formally defined and typed based on the representation
formalism used, and relationships between each defined piece of knowledge explicitly
determined. This is a time consuming activity and requires the cooperation of various domain
experts [10]. As so much time and effort is invested in developing computerized guidelines, it
is important that they can be shared between institutions and integrated into different systems,
which in itself, brings on additional challenges. The formalization process can also cause
variations in guideline interpretation, leaving room for potential errors of omission and biases
[10].
The purpose of our research was twofold. Our first goal was to identify tools that have been and
are currently being used for the representation and execution of guidelines and guideline like
documents, with particular attention to free and open source systems. Our choice to focus on
tools with free and open source licenses was motivated by their advantage over proprietary
software, especially for use within education and research. Free and open source tools can be
used at no cost, and can be freely distributed. Additionally users are given access to the source
code, which gives them the option to modify it in order to make improvements or adapt the tool
to their needs. The identified tools were classified into toolchains comprising of guideline
modeling and execution tools related to a specific representation formalism. Subsequently we
determined their availability, license and critical features such as research group, source of
funding, medical domains in which tools were used, and ascertained their applicability based
on objective evaluation criteria derived from literature. Our second goal was to evaluate freely
available tools, especially free and open source tools, using the aforementioned criteria,
determining their comparative advantages and disadvantages.
8 |
P a g e
Our research questions are presented below:
1. What tools are used for the representation and execution of guidelines and guideline like
documents, with particular attention to freely available systems, especially free and
open source tools.
2. What are the comparative advantages and disadvantages of at least two of these tools,
in creating an initial guideline representation and updating it with a new version of the
same guideline? Based on experimental results, how can these tools be improved?
3. What tools should be recommended for teaching purposes, research and adoption as a
standard for guideline implementation in clinical practice?
In the following chapter, we report the results of our literature review. Chapter 3 discusses
evaluation criteria derived from literature. In Chapter 4 we evaluate two freely available
systems in their function to represent and execute a portion of a clinical guideline. Chapter 5
presents our general discussion and conclusion.
9 |
P a g e
2
LITERATURE REVIEW
Introduction
Knowledge representation is defined as ‘the application of logic and an ontology in order to
construct computable models of some domain’[153]. Knowledge representation methods and
tools enable the transformation of text based guidelines into computerized format, making their
tacit knowledge explicit. Formal and expressive representation models provide an in-depth
understanding of clinical procedures, and a precise and unambiguous description of the
guideline itself [140]. To facilitate the implementation of guidelines into clinical practice and
provision of point of care recommendations, the knowledge that has been represented in
computerized format has to be interpreted by a computer program known as a parser, then
applied as an action against patient data. This is known as knowledge execution.
Following the introduction of Arden Syntax in 1989, one of the first methods used to represent
guideline knowledge, research into guideline representation and execution methods and tools
gained momentum in the nineties and early 2000’s. Various research groups developed their
tools based on five main approaches, namely rule-based, document-centric, expression
languages, model-centric, and decision trees. In the rule based approach, formalizations use sets
of if-then statements to create assertions, as well as rules on how to act upon these
assertions[11]. Examples include Arden Syntax [12] and the more recent guideline definition
language (GDL) [13]. Document-centric approaches organize heterogenous information from
text based guidelines into formal models by identifying and tagging components relevant to
their operationalization [14]. GEM [15] is one of the more known document centric
approaches. Approaches using expression languages such as GELLO [16] formalize guidelines
as standard queries and expressions for decision support. In model-centric approaches,
conceptual models (task network models) are used to represent guidelines as hierarchical
structures that contain networks of actions and decisions that unfold over time [17]. Generated
models visually range from simple flowcharts like in GLIF [18] to three dimensional
metaphoric representations that use familiar day to day objects to represent different aspects of
a guideline like in Asbru [19]. The last approach comprises of decision trees, models augmented
with probabilities and utilities to provide an optimal action strategy [9]. They take into
consideration the uncertainty of the outcome of selected treatments by specifying prior
probabilities in the population for different outcomes, and the decision-maker’s preferences by
specifying their utilities for different outcome states [9]. An example of this approach is the
EsPeR system [20]. Fuzzy Cognitive Maps (FCMs) are similar to decision trees. In addition to
10 |
P a g e
representing guideline knowledge, they enable reasoning with causality and uncertainty. An
example of its use can be seen in [21].
To implement these approaches, a number of tools have been developed under either free and
open source or proprietary licenses. Free and open source licenses allow users to run software
for any purpose, modify original source code, and freely redistribute original or modified copies
[154]. Proprietary software on the other hand, keeps its source code hidden, and does not allow
any modifications or free redistribution.
There is no recent complete overview of guideline representation and execution tools in
literature. In the first part of our study, we address this problem and answer our first research
question through a literature review. Particular attention is given to freely available tools,
especially free and open source systems.
Methods
A comprehensive search of peer review journals, including conference papers and reports, was
conducted using Medline and Embase databases. The search consisted of the following
keywords and their synonyms: ((((<knowledge representation> or <knowledge execution>) and
<guidelines>) or <computerized guidelines>) or (<open source> and <decision support>)). The
search was limited to English language papers with no restrictions on date of publication. The
final search was performed on February 22, 2017 and is shown in Appendix A.
Title and abstract screening, full text screening, and data extraction were performed by one
researcher (M.G.). A second researcher (S.M.) was consulted when a decision could not be
reached after a full text review. A third reviewer was to be consulted in case of disagreement.
Additional papers were selected through a bibliographic search.
The inclusion and exclusion criteria of our literature review were decided by 3 researchers
(M.G., S.M., D.S.) and are presented below.
Inclusion criteria:
1. Papers that describe or mention the use of a tool or a framework for guideline
representation and/or execution.
2. Papers that present specifically developed guideline based decision support systems
built atop general purpose tools such as Protégé and inference or rule engines.
3. Papers that describe the development of a guideline based decision support system built
using tools for guideline representation and/or execution.
11 |
P a g e
4. Papers that define evaluation criteria or design criteria for guideline representation and
execution tools or systems
2.
Exclusion criteria:
1. Literature reviews and syntheses.
2. Papers that describe a methodology without mention of the tools used.
3. Papers on proposed languages with no tools developed at the time of writing.
4. Papers that describe decision support systems built using general purpose tools such as
C++, Prolog, CLIPS, etc.
5. Papers that describe tools for building guideline based decision support systems that are
used for purposes other than knowledge representation and execution.
6. Papers describing representation and execution of knowledge other than guidelines such
as care plans, clinical pathways, or knowledge in general.
7. Papers on perception of use of methods and tools.
8. Unpublished papers such as theses.
The resulting papers were grouped into toolchains, and the following information was
extracted:
1. Title, author, date of publication.
2. Name of tool with description of its functionality.
3. Institute or department (including country) where research was done, as well as source
of funding for the research.
4. Focus of research – development of a tool for guideline representation and/or execution
versus development of a guideline based decision support system.
5. Medical domain in which the tool or decision support system was used.
6. Number of patient records used for validating the tool or decision support system.
7. Use of Protégé as an authoring tool in the framework.
8. Possible evaluation criteria.
Results
206 papers were reviewed in full by the first researcher(M.G.), 22 by the second researcher
(S.M.), and as there was no disagreement between the first and second researcher, the third
researcher was not consulted. In total 131 papers identifying 19 toolchains were included.
Majority of the papers presented tools originating from one toolchain. In two papers, more than
one toolchain was discussed. The first paper was a case study that compared and evaluated the
formalization of a guideline using six different methodologies and tools [17]. The second paper
described the development of a decision support system built atop two different toolchains [22].
Figure 1 presents the results of our article screening and inclusion.
12 |
P a g e
Figure 1: Screening Flow Chart
Table 1 summarizes our literature review results. In the first three columns we present basic
information that includes the name of the tool chain, the year the representation formalism the
toolchain is based on was introduced, and the name of the institute that developed the
formalism. The next two columns present the number of papers referencing a specific toolchain
in our literature review, as well as the year of the most recent publication to help determine if
research on a specific toolchain is ongoing. The next column indicates if Protégé was used as
an authoring tool within a specific toolchain. Protégé is a general purpose knowledge authoring
tool that has been adapted for use by different groups to represent guidelines using different
formalisms. In our last column we identify the domains a toolchain has been tested or used in.
A brief description of each toolchain follows.
13 |
P a g e
Table 1: Literature Review Results
Toolchain Year Introduced
Founding Institute(s) Last Paper Paper Count Protégé as Authoring Tool
Domain(s) in which toolchain has been applied/tested
Arden Syntax 1989 Columbia Presbyterian Medical Center & IBM, USA
2016 6 No DVT, Lyme Borreliosis, Obstetrics, Screening and Preventative guidelines
PROforma 1992 Advanced Computation Laboratory of Cancer Research, UK
2014 10 No Acute Asthma, ALL, Breast Cancer, CHD, Chronic Cough, Dyspepsia, HTN, IBS, Thyroid Nodules Prestige 1995 Collaboration of 30 organizations
from 8 countries
2000 3 No Angina, Diabetes
EON 1996 Stanford University, USA 2003 7 Yes Asthma, Breast Cancer, Chronic Cough, HIV, HTN, Influenza Vaccine
PRODIGY 1996 University of Newcastle upon Tyne, UK
2004 5 Yes Asthma, Chronic Cough, HTN, Stable Angina GASTON 1997 Eindhoven Technical University &
Maastricht University, The Netherlands
2009 8 Yes Anesthesia, Diabetes, HF, HTN, Leukemia, Psychiatry GLARE 1997 University of Eastern Piedmont &
Azienda Ospedaliera S. Giovanni Battista Hospital, Italy
2013 14 No Bladder Cancer, HF, Management of harmful drinking and alcohol dependence, Non-Hodgkin lymphoma, Reflux
esophagitis, Severe Trauma Management, Stroke Asbru 1998 Vienna University of Technology,
Austria & Stanford University, USA
2009 10 No Breast Cancer, Chronic Cough, Diabetes, HTN, ICU, Jaundice, Mechanical Ventilation of Newborn Infants GLIF 1998 InterMed Collaboratory: Harvard,
Columbia & Stanford Universities, USA
2014 22 Yes Acute Stroke, Acute MI, Angina, Cancer (Breast, Colorectal), CHF, Chronic Cough, Chronic Hepatitis C,
Depression, Dermatology, Diabetes, DTP, Influenza Vaccination, Hypercholesterolemia, Hyperkalemia
screening, HTN, LBP, Sleep Apnea, Pulmonary Embolism, Thyroid screening, Tobacco Dependence GUIDE/
NEWGUIDE
1998 University of Pavia, Italy 2005 9 No Acute Stroke, AML, Breast Cancer, Chronic Cough, HF, HTN, Pressure Ulcer
GEM 2000 Yale University School of Medicine, USA
2010 4 No Extravasation of infused medication, Smoking cessation DeGeL 2001 Ben Gurion University in Beer
Sheva, Israel
2016 14 No Cardiology, CHF, COPD, Endocrinology (Primary Hypothyroidism), Gynecology, HF, Hypercholesterolemia,
HTN, PID, Preeclampsia, Pulmonology, Toxemia Stepper 2001 University of Economics, Czech
Republic
2004 3 No Unknown SAGE 2002 IDX Corporation, Stanford
University, Mayo Clinic Rochester, University of Nebraska, Intermountain Health Care &
Apelon Inc., USA
2015 10 Yes Acute Stroke, CAP, Diabetes, Headache, HTN, Immunization, Lab Alerting, Pressure Ulcer
HELEN 2004 Heidelberg University Medical Center, Germany
2004 1 Yes Apnea (newborns), Hyperbilirubinemia, Small Cell Lung Cancer, Uveitis
Semantic Web Technologies
N/A Various 2016 8 Yes Asthma, ATP, Breast cancer, HF, HTN, Imaging (RPG), Nursing, Ocular Conditions, Pediatric Stoma, PGx guideline, Pressure Ulcers, Stroke, UTI, Upper RTI
OTHER
SIEGFRIED Unknown Duke University Medical Center, USA
1998 2 No Low back pain HGML
Markup Tool
Unknown Rutgers University, USA 2000 1 No Stroke Medical Text
Markup Tool
Unknown University of Pavia, Italy 2003 1 No Hypertension
Arden Syntax
Overview: Arden Syntax is a well-established rule based formalism used to represent medical
knowledge in the form of medical logic modules (MLM’s), where each one houses the
knowledge needed to make one clinical decision. It enables knowledge sharing between
institutions through the use of ‘curly braces’, which separate parts of code specific to a local
information system.
Arden Syntax was first introduced in 1989, and was declared a standard by
Health Level 7 (HLA7) in 1992. It was initially developed to implement simple clinical alerts,
14 |
P a g e
reminders and recommendations, but has been used for the implementation of clinical practice
guidelines as well.
Tools: Arden2ByteCode [23], ArdenSuite [23, 27]. Both tools are currently available.
Arden2ByteCode is a free and open source tool downloadable from GitHub. ArdenSuite is
available under license from www.medexter.com with an offer of a 30 day free trial.
Decision Support Systems: EGADSS [23], HELP [26], NEXPERT Object [24, 25], Arden
Syntax GBDSS for Lyme Borreliosis and Obstetrics [3, 27].
PROforma
Overview: PROforma was introduced in 1992 by the Advanced Computation Laboratory of
Cancer Research, UK, as a formalism that combines logic programming and object oriented
modeling [17]. It consists of a graphical notation for knowledge design and a formal
representation language that enables execution. PROforma enables the specification of
guidelines in terms of tasks and collection of tasks, with each task defined as being either a
plan, decision, action or an enquiry [28, 29, 30, 31]. PROforma has been used in building
systems for clinical task management and typically presents an authoring system for
formalizing knowledge, and a software engine for enacting the tasks in a clinical setting [31].
Tools: PROforma graphical knowledge editor [29, 30], PROforma engine [29, 30, 32], Arezzo
[28, 31, 35], Tallis [17, 22, 31, 35]. Arezzo is available under commercial license, whereas
Tallis is available upon request from COSSAC
3.
Decision Support Systems: GBDSS (monitoring and assessing of hypertensive patients by
pharmacists) [35], COGENT [33], HeCaSe2 [34]. Non-guideline based decision support
systems developed using PROforma include CAPSULE, RAGS, CADMIUM, Arno System,
RetroGram, ERA, and LISA [28, 31].
Prestige
Overview: The Prestige project (Guidelines in Healthcare) was a collaboration of about 30
organizations from eight countries started in December of 1995 [36]. Its focus was on the
application of telematics technologies for the dissemination and implementation of clinical
practice guidelines and protocols. The project led to the creation of several clinical applications
that were linked to established IT environments, with programs for validation for use in clinical
practice. One of the aims of the project included the creation of a shared methodology for
conversion of text based clinical guidelines into a formal machine readable structure that could
be integrated with computerized clinical data to generate recommendations for action. Another
aim was to develop a sound and sustainable process for managing, disseminating and locally
adapting computerized versions of clinical guidelines. Each application created within Prestige
contained a common Protocol Manager, a software component to identify clinical guideline
15 |
P a g e
recommendations relevant to a specific patient, with interfaces to connect it to the applications
operational front end, patient record, and knowledge server [36].
Tools: GAUDI [36, 155], GLEAM [36], PAT [36]
Decision Support Systems: Prototype GBDSS for the management of diabetes [37]
EON
Overview: EON, developed in 1996 at Stanford University, provided a suite of models and
software components that were used to develop guideline based decision support systems [17].
Research on EON ended in 2003 and its results were carried over to the SAGE project.
Eligibility criteria, definitions of abstractions, guideline algorithms, decision models, and
recommended actions were structured within the Dharma guideline model. The execution
system obtained patient data through the Tzolkin temporal data mediator or from user input in
order to generate situation specific recommendations. WOZ provided explanation services for
other components.
Tools: EON uses Protégé 2000 for authoring guidelines. Padda execution server generates
situation specific recommendations based on patient data [17, 39, 40, 41, 43].
Decision Support Systems: ATHENA [38, 40, 42], T-HELPER [43].
PRODIGY
Overview: PRODIGY (Prescribing RatiOnally with Decision Support In General Practice
study) was introduced in 1996 by the University of Newcastle upon Tyne, UK. It was a
guideline based decision support system, developed in several phases, and used mainly in
chronic disease management. Phase I and II were implemented as modules by system vendors,
extending their existing EPR systems and accessing the coded information to help direct choices
in the guidelines [44, 45]. Phase II was introduced for use in acute disease management. Phase
III focused on the integration of the system with legacy systems through the introduction of
standardized interfaces, terminology mapping and numeric unit conversion knowledge bases
[46]. PRODIGY structured guidelines as a set of choices (patient scenarios) that were to be
made by the physician, and enabled the synchronization of patient management with respective
guideline recommendations.
Tools: PRODIGY used Protégé as its authoring tool [17, 44, 45]; KWIZ [47]
Decision Support Systems: PRODIGY was integrated with two primary care clinical systems
in the UK [46].
GASTON
Overview: GASTON was introduced in 1997, developed jointly by Eindhoven Technical
University and Maastricht University in the Netherlands. The framework consists of a
formalism that uses primitives (based on version 2.0 of GLIF), problem solving methods and
ontologies, for the purpose of representing guidelines varying in complexity and granularity, as
well as in different application domains. GASTON was later extended to include the use of
16 |
P a g e
intentions, as they made the developed systems more flexible [48]. A guideline authoring
environment allowed authors to define guidelines using the developed representation
formalism, and an execution environment translated the defined guidelines to a symbol level
representation that could be read and processed by an execution engine [50].
Tools: GASTON framework consists of a Protégé based editor and an execution engine [48,
50, 55]; GASTINE (GASTon INtentional Expressions) is built atop GASTON and focuses on
intention based guidelines [54].
Decision Support Systems: CritICIS [49, 50, 51, 52, 53, 54], GRIF [49, 50, 51],
Multidisciplinary psychoactive drug selection advisor system (M-PADS) [49, 50, 51],
TANDEM project [50, 51], Medical Guideline Technology (MGT) project [50, 51].
GLARE
Overview: GLARE is a domain independent system used to acquire, represent and execute
clinical guidelines. The system was introduced in 1997 in a long-term cooperation between the
Dipartimento di Informatica of Università del Piemonte Orientale, Alessandria, Italy, and
Azienda Ospedaliera S. Giovanni Battista, Torino, one of the largest hospitals in Italy. It
presents a limited set of clear representation primitives that cover most of the relevant aspects
of a guideline, and provides facilities to deal with its context dependent character, taking into
account resource availability, times and cost. It allows users to deal with the temporal aspects
of guidelines and is able to automatically detect different forms of syntactic and semantic
inconsistencies in the guidelines being acquired. The execution module also incorporates a
decision support facility which allows users to compare alternative paths while navigating
through the guideline. Over the years GLARE has evolved to provide users with extended
functionalities such as managing of exceptions, enabling semi-automatic adaptation to specific
execution contexts, automatic treatment of temporal constraints, decision making based on
decision theory, model based verification, and execution in a distributed environment.
Tools: GLARE comprises of the following modules: CG_KRM (Clinical Guidelines
Knowledge Representation Manager) [58, 59, 62, 63, 64, 160], CG_AM (Clinical Guidelines
Acquisition Manager) [56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 160], CG_EM (Clinical
Guidelines Execution Module) [56, 57, 60, 61, 62, 63, 64, 65, 66, 67, 68, 160], CG_IM (Clinical
Guidelines Interface Module) [56, 63, 64], CG_SS (Clinical Guidelines Support System [58,
59], CG_ES (Clinical Guidelines Evaluation System) [58, 59].
Asbru
Overview: Asbru was developed in 1998 within the Asgaard project by researchers at the
Vienna University of Technology and Stanford University Medical Informatics department.
Asbru is a time oriented and intention based language defined in Backus-Naur form (BNF),
used to represent clinical guidelines and protocols as time oriented skeletal plans [70]. A
hybrid-Asbru representation format is used within the DeGeL framework.
17 |
P a g e
Tools: AsbruView
[17, 19, 69, 70], Asbru Knowledge Acquisition Tool [70, 71], Asbru
Interpreter [72, 75], Guideline Markup Tool [69, 73, 74], DELT/A [71, 76]. AsbruView, Asbru
Interpreter and DELT/A are freely available for download.
GLIF
Overview: GLIF was introduced in 1998 by the InterMed Collaboratory, a collaboration of
medical informatics departments at Harvard, Columbia and Stanford Universities, with the
support of the National Library of Medicine [17, 18, 81, 88, 89]. Its purpose was to create a
representation model that could be shared among different institutions and software systems,
and to bring together the most useful features of other guideline models while incorporating
health care standards. GLIF specifies an object oriented model for guideline representation and
is essentially a flowchart that represents a temporally ordered sequence of steps [77]. It
encourages a top down process of modeling and supports three levels of abstractions: level 1 is
conceptual, a human readable flowchart of decisions and actions; level 2 is a computable
specification that can be verified for logical consistency and completeness; level 3 is an
implementable specification that has information needed for local adaptation, as well as for
mapping the guideline onto the patient record [18, 82].
Tools: GLIF uses Protégé as an authoring tool [17, 18, 22, 82]; Partners Computerized
Algorithm Processor and Editor (P-CAPE) [85, 90], DSG tool suite [77, 81, 82], GLEE [18, 79,
87, 88, 89], PRESGUID project [80], OCL compliant GELLO engine [16].
Decision Support Systems: Pilot execution engine integrated with MUltimedia Distributed
Record version 2 [83], MEKRES [78, 161], GC3 [84], ASTI [91], EsPeR [158].
GUIDE/NEWGUIDE
Overview: GUIDE was proposed by the University of Pavia in 1998 as a web distributed
framework designed to manage the development and implementation of guidelines, taking
patient and organizational preferences into consideration [94, 96]. The framework initially
consisted of a formal graphical representation language augmented with the possibility of
linking decision analytic models, a front end for accessing the patient database, and an interface
to the organizational model of the practice [94, 96]. The formalized guideline was linked to the
patient record enabling an inference engine to provide real time patient specific advice upon
request of the user. A library of guidelines was also maintained by GUIDE. NewGuide evolved
from GUIDE. It introduced a Virtual Medical Record (vMR), a logging system that allowed the
tracing of health care process details, and a repository for storage of formalized guidelines [93,
94].
Tools: GUIDE consists of an editor [17, 93, 94, 95, 96], inference engine [92, 95] or GET
(Guideline Enactment Tool) [94].
Decision Support Systems: Decision support system for ulcer prevention [97, 98], careflow
18 |
P a g e
GEM
Overview: GEM, developed in 2000, was derived from previous work with augmented decision
tables done at Yale University School of Medicine [15, 100]. It uses a document centric
approach to define an XML based model that stores and organizes heterogenous information
contained in clinical guidelines. GEM models the names of guideline elements and their
attributes, element contents, and document structure, and defines them into a multi-level
hierarchy, which consists of more than 100 discrete elements in 9 major branches. GEM became
an ASTM standard in 2006. In [100] the authors discuss their efforts in translating GEM II from
XML into OWL, a semantic web ontology language, in order to enable wider integration.
Tools: GEM Cutter [15, 100, 159]. The tool is freely available for download.
Decision Support Systems: Decision support system for smoking cessation [101]
DeGeL (Digital electronic Guideline Library)
Overview: DeGeL was introduced in 2001 at the Ben Gurion University in Beer Sheva, Israel.
The goal was to create and implement a distributed web based modular architecture that enabled
gradual conversion of clinical guidelines from text to a fully structured machine comprehensible
representation, passing through semi structured and semi-formal stages [107]. Asbru was used
for the formal representation, however other languages such as GEM could be used as well. In
[108, 109] there is also mention of the intention to use GLIF ontology. Besides mark up and
run time application tools, DeGeL has offered a variety of accessory tools. These include:
Vayduria, a search, retrieval and visualization tool; an eligibility determination module;
Qualiguide, a retrospective quality assessment tool; IndexiGuide for guideline classification;
and VisiGuide browsing tool. DeGeL offers Idan, an architecture that allows access to any
heterogeneous medical database so as to query raw clinical data and its abstractions. KNAVE
II intelligent visualization and exploration client uses Idan’s computational capabilities to
display and explore the patient's raw data or derived concepts. MEIDA system includes a
vocabulary server and a search engine and enables search and retrieval of standard terms
defined using controlled medical vocabularies. DeGeL.NET is a new version of the digital
library and its different tools. It also uses Asbru ontology, and its main modules include a
knowledge base server, GESHER as the guideline-specification tool, and Spock for runtime
application of clinical guidelines. Other modules include a guideline database that supports the
hybrid multiple ontology representation; a module responsible for guideline-knowledge
creation, reading, updating, and deletion; DeGeLook, a new guideline search engine for
enhanced guidelines retrieval; DeGeLock, an authorization and authentication module for
group-based authorizations; and a web-service API that enables the guideline knowledge-base
server to accept client requests and perform requested transactions.
Tools: Uruz [105, 106, 107, 108], Gesher [102, 103, 104, 110, 112], Spock [102, 103, 106, 108,
109, 113, 114, 115]. Gesher and Spock are available for licensed use.
19 |
P a g e
Stepper
Overview:
Stepper is a markup tool developed in 2001, that supports step by step formalization
of clinical guidelines. The tool transforms guideline text with increasing level of detail,
rearranges it into an XML knowledge base, and eventually exports it into an operational
representation. Stepper comprises of an embedded XSLT processor, which is responsible for
the non-interactive part of the transformation, while markup and interactive transformations are
carried out via rules expressed in XKBT language. The tool and methodology were created with
the purpose of guiding the formalization process, at the same time reducing the amount of
involvement of the domain expert and minimizing information loss. Stepper is freely available
for download [116, 117, 118].
SAGE
Overview: SAGE (Standards-Based Sharable Active Guideline Environment) project was
started in 2002 as a collaboration between IDX Corporation, Stanford Medical Informatics,
Mayo Clinic (Rochester), the University of Nebraska, Intermountain Health Care, and Apelon
Inc., with the aim of providing a shared guideline model which could be integrated with vendor
systems for decision enhancement, without replacing vendor software functionality. To ensure
interoperability with vendor systems, SAGE employs a common layer of standard information
models (patient data, organizational and workflow models), and terminologies. According to
[124], SAGE model derived elements from PROforma, GLIF, EON, and PRODIGY. SAGE
has been used in tandem with rule and inference engines such as uEngine [7, 121, 126].
Tools: SAGE uses Protégé as its authoring tool [7, 119, 120, 122, 123, 124]; KWIZ [47, 119],
SAGEDesktop [119], SAGE execution engine [119, 123].
Decision Support Systems: GBDSS for headache diagnosis [125].
HELEN
Overview: In 2004, the Department of Medical Informatics at the Heidelberg University
Medical Center in Germany introduced a modular framework called HELEN, that handled both
guideline authoring and execution. Its purpose was to implement clinical guidelines for the
Department of Neonatology [127].
Tools: Comprises of a Protégé based guideline editor, guideline viewer for web based browsing,
engine for executing encoded guidelines [127]. The framework is freely available for download
from the University’s web page.
Semantic Web Technologies and Inference Engines
Overview: Recent research indicates a preference for using semantic web technologies for
guideline representation, and inference engines for their execution. A number of decision
support systems have been developed and are listed below.
Tools: HeD editor (authoring tool) [133], OWL 1 DL, OWL 2 DL and OWL 2 DL + SWRL
20 |
P a g e
Decision Support Systems: CPG-EX (Protégé and JENA) [128,131]; Decision support system
for nursing clinical practice guidelines (Protégé and JENA) [130]; Decision support system
prototype (Protégé and JESS) [129]; FCM-uUTI DSS (Protégé and EYE) [21, 134]. Protégé, as
well as rule engines JENA and EYE, are free and open source tools.
Other
SIEGFRIED (System for Interactive Electronic Guidelines with Feedback and Resources for
Instructional and Educational Development) was a web based system that was developed to
integrate clinical guidelines into clinical flow by interactively presenting them at the point of
care. It elicited clinical data from the physician and provided appropriate recommendations
based on the underlying clinical guideline. To facilitate the entry of content and logic, the
Guideline Entry Wizard, a tool modeled after the wizard concept in PC based applications, was
developed [135, 157].
HGML Markup Tool was first presented at the 2000 AMIA Annual Symposium by the
Department of Computer Science at Rutgers University in New Jersey. Its aim was to present
a guideline delivery system that allowed a clinician to extract relevant content easily. It enabled
existing guidelines to be manipulated and viewed in different formats, at various levels of detail
based on the needs of the user, while preserving their original published format [136].
Medical Text Markup Tool was an experimental tool developed for the modularized
representation of clinical guidelines. Through text markup, hierarchical modules, which could
be part of a larger module or have other modules as their parts, were formed. Portions of the
guideline were copied and pasted into an XML editor, thereby creating XML files. MTM tool
complemented guideline execution tools such as GUIDE by determining which tasks could be
represented and which portions of text were related to a particular task [137].
Table 2 presents a summary of the toolchains with their main features. For a more detailed
description see Appendix B and C.
21 |
P a g e
Table 2: Toolchains
Author(s)
System
Function
Institute
Arden Syntax
Lam, 1993 HELP system GBDSS University of Utah, UT, USA Jenders and Barnett, 1993 NEXPERT OBJECT Editor and inference engine Massachusetts General Hospital, MA, USA
Jenders et al., 1994 NEXPERT OBJECT Editor and inference engine Massachusetts General Hospital, MA, USA Gietzelt et al., 2012 Arden2ByteCode
Authoring and Execution University of Braunschweig & Hannover Medical School, Germany
Seitinger et al., 2014 GBDSS for Lyme
borreliosis; ArdenSuite GBDSS; Execution engine
Medical University of Vienna, Medical University Graz, & Medexter Healthcare GmbH, Austria Seitinger et al., 2016 GBDSS for Lyme
borreliosis, Obstetrics; ArdenSuite
GBDSS; Execution engine Medexter Healthcare GmbH; Medical University of Vienna; Vienna Hospital Association, Austria
PROFORMA
Fox et al., 1997 PROforma editor &
enactment engine Authoring and Execution
Imperial Cancer Research Fund & North End Medical Centre, UK
Fox et al., 1998 PROforma editor & enactment engine Authoring and Execution Imperial Cancer Research Fund, UK Bury et al., 2000 Arezzo Authoring and Execution Imperial Cancer Research Fund, UK Bury et al., 2001 PROforma execution engine Execution engine Imperial Cancer Research Fund, UK Peleg et al., 2003 Arezzo Authoring (case study) Cancer Research, UK (PROforma part) Sutton and Fox, 2003 Arezzo, Tallis Authoring and Execution Oxford Brookes University, UK
Sutton et al., 2006 GBDSS using Tallis Authoring and Execution Oxford Brookes University, UK Grando et al., 2012 COGENT Authoring and Execution University of California, San Diego, CA
Isern et al., 2012 HeCaSe2/Tallis Authoring and Execution Rovira i Virgili University; ITAKA Research Group, Spain
Peleg et al., 2014 Tallis (Protégé)
Authoring and Execution (Tallis part)
University of Haifa, Israel; Oxford University, UK; University College London, UK; Whittington Health
NHS Trust, UK; Deontics Ltd., UK; Harvard Medical School, MA, USA; Mayo Clinic College of Medicine, MN, USA; Albano and University la Sapienza, Italy;
University of Leipzig, Leipzig, Germany; Endocrinology Associates, AR, USA; Arcispedale Santa
Maria Nuova, Italy; Odense University Hospital, Denmark
PRESTIGE
Gordon et al., 1997 GAUDI Authoring
Royal Brompton & Harefield NHS Trust, UK; NHS Information Management Centre, UK; Sowerby Unit for
Primary Care Informatics, University of Newcastle, UK Gordon and Veloso, 1999 GAUDI, GLEAM Authoring Royal Brompton & Harefield NHS Trust, UK
Barahona et al., 2000 GBDSS (Diabetes) GBDSS Nova University of Lisbon and UNINOVA, Portugal
EON
Tu and Musen, 1996 T-HELPER system GBDSS Stanford University School of Medicine, CA, USA Tu and Musen, 1999 EON (Protégé based) Authoring Stanford University School of Medicine, CA, USA Tu and Musen, 2000 Dharmapadda Authoring and Execution Stanford University School of Medicine, CA, USA Goldstein et al., 2000 ATHENA GBDSS Stanford University School of Medicine, CA, USA
Tu and Musen, 2000 EON (Protégé based)/
Dharmapadda Authoring and Execution Stanford University School of Medicine, CA, USA Tu and Musen, 2001 Dharmapadda Authoring and Execution Stanford University School of Medicine, CA, USA Peleg et al., 2003 EON (Protégé based) Authoring (case study) Stanford University School of Medicine, CA, USA
PRODIGY
Johnson et al., 1999 PRODIGY Authoring and Execution University of Newcastle upon Tyne, UK; Stanford University School of Medicine, CA, USA Johnson et al., 2000 PRODIGY Authoring and Execution University of Newcastle upon Tyne, UK Johnson et al., 2001 PRODIGY 3 Authoring and Execution University of Newcastle, UK; Stanford University
School of Medicine, CA, US Peleg et al., 2003 PRODIGY Authoring (case study) University of Newcastle, UK (PRODIGY part) Shankar et al., 2004 KWIZ Authoring University of Newcastle, UK; Stanford University
School of Medicine, CA, US
GASTON
De Clercq et al., 1999 CritICIS knowledge base
editor/GASTON Authoring
Eindhoven University of Technology, University of Maastricht, Catharina Hospital, Eindhoven, NL De Clercq, 2000 GASTON Authoring and Execution Eindhoven University of Technology, NL De Clercq et al., 2000 GASTON Authoring and Execution University of Maastricht, NL
De Clercq et al., 2001 GASTON Authoring and Execution University of Maastricht, Eindhoven University of Technology, Catharina Hospital, Eindhoven, NL De Clercq et al., 2001 GASTON Authoring and Execution University of Maastricht, NL Latoszek-Berendsen et al.,
22 |
P a g e
Latoszek-Berendsen et al.,
2007 GASTON Authoring and Execution
University of Maastricht, MEDECS, Eindhoven, University of Amsterdam, NL Latoszek-Berendsen et al.,
2009 GASTINE Authoring and Execution University of Maastricht, NL
GLARE
Guarnero et al., 1998 CGM Authoring and Execution Laboratorio di Informatica Clinica, Italy. Terenziani et al., 2000 CGM Authoring and Execution University of Eastern Piedmont, Italy Terenziani et al., 2001 CGM Authoring and Execution University of Eastern Piedmont, Italy Terenziani et al., 2002 GLARE Authoring and Execution University of Eastern Piedmont, Italy Terenziani et al., 2003 GLARE Authoring and Execution University of Eastern Piedmont, Italy Terenziani et al., 2004 GLARE Authoring and Execution University of Eastern Piedmont, Italy Terenziani et al., 2004 GLARE Authoring and Execution University of Eastern Piedmont, Italy Molino et al., 2006 GLARE Authoring and Execution University of Eastern Piedmont, Italy Giordano et al., 2006 GLARE Authoring and Execution University of Eastern Piedmont, Italy Bottrighi et al., 2006 GLARE Authoring and Execution University of Eastern Piedmont, Italy Terenziani et al., 2007 GLARE Authoring and Execution University of Eastern Piedmont, Italy Terenziani et al., 2008 GLARE Authoring and Execution University of Eastern Piedmont, Italy Leonardi et al., 2012 GLARE Authoring and Execution University of Eastern Piedmont, Italy Bottrighi et al., 2013 GLARE Authoring and Execution University of Eastern Piedmont, Italy
Asbru
Miksch et al., 1998 AsbruView Authoring Vienna University of Technology, Austria; Stanford University, CA, USA Shahar et al., 1998
Asbru knowledge-acquisition tool; Asbru
Interpreter
Authoring and Execution Stanford University, CA, USA Kosara and Miksch, 2001 AsbruView Authoring Vienna University of Technology, Austria
Kosara et al., 2002 Guideline Markup Tool Mark-up tool Vienna University of Technology, Austria Peleg et al., 2003
AsbruView
GLIF, EON, PRODIGY,
Arezzo, GUIDE
Authoring Vienna University of Technology, Austria (Asbru part) Votruba et al., 2004 Guideline Markup Tool Mark-up tool Vienna University of Technology, Austria Votruba et al., 2004 Guideline Markup Tool Mark-up tool Vienna University of Technology, Austria Votruba et al., 2008 Asbru Interpreter Execution engine Vienna University of Technology, Austria Seyfang et al., 2009 DELT/A Mark-up tool Vienna University of Technology, Medical University of
Vienna, & Danube University Krems, Austria
GLIF
Zielstorff et al., 1998 P-CAPE Authoring and Execution Partners HealthCare System, MA, USA Ohno-Machado et al., 1998 Evaluation Criteria Evaluation Criteria
Brigham and Women's Hospital, MA, USA; Stanford University, CA, USA; Massachusetts General
Hospital, MA, USA; Columbia University, NY, USA; BBN Systems and Technologies, MA, USA;
DSG, MA, USA
Greenes et al., 1999 DSG's tool suite Authoring and Execution DSG Harvard Medical School, MA, USA Boxwala et al., 1999 DSG's tool suite Authoring and Execution DSG Harvard Medical School, MA, USA
Peleg et al., 2001 Evaluation Criteria Evaluation Criteria Stanford University School of Medicine, CA, USA Peleg et al., 2001 DSG's tool suite/Protégé Authoring and Execution Stanford University School of Medicine, CA; DSG
Harvard Medical School, MA, USA Seroussi et al., 2001 ASTI GBDSS Service d'Informatique Medicale, France.
Gillois et al, 2001 EsPeR GBDSS
SPIM, Faculte de Medecine France; Laboratoire SPIEAO, Faculte de Medecine Nancy I, France, LERTIM, Faculte de Medecine, Marseille, France Wang and Shortliffe, 2002 GLEE Execution engine Columbia University, NY, USA
Dufour et al., 2003 PRESGUID project Authoring and Execution LERTIM University of the Mediterranean, France Maviglia et al., 2003 P-CAPE Authoring and Execution Partners Healthcare System, Inc., MA; Harvard Medical
School, MA; Healthvision, Inc., MA, USA Peleg et al., 2003 GLIF (Protégé based) Authoring (case study) DSG Harvard Medical School, MA; Columbia
University, NY, USA Wang et al., 2004 GLEE Execution engine
Columbia University, NY; Stanford University, CA; DSG Harvard Medical School, MA; Eclipsys Corporation, MA, USA; University of Haifa, Israel Boxwala et al., 2004 Protégé/GLEE Authoring and Execution
DSG Harvard Medical School, MA; Stanford University; Columbia University, NY; Eclipsys Corporation, MA, USA; University of Haifa, Israel Kolesa et al., 2005 MUDR2 system Execution engine EuroMISE Center and Center for Biomedical
Informatics, Czech Republic Choi et al., 2007 GLEE Execution engine
Spaulding Rehabilitation Hospital, MA; Columbia University, NY; University of Rochester, NY,
USA
Buchtela et al., 2008 GLIF editor/MEKRES Authoring EuroMISE Center and Center for Biomedical Informatics, Czech Republic
23 |
P a g e
Buchtela et al., 2008 MEKRES Authoring EuroMISE Center and Center for Biomedical Informatics, Czech Republic Peleg et al., 2008 GLEE Execution engine University of Haifa, Israel
Mei et al., 2011 GELLO engine Execution engine IBM Research, China Liu et al., 2012 GELLO engine Execution engine IBM Research, China
Peleg et al., 2014 Protégé (Tallis) Authoring (GLIF part)
University of Haifa, Israel; Oxford University, UK; University College London, UK; Whittington Health
NHS Trust, UK; Deontics Ltd., UK; Harvard Medical School, MA, USA; Mayo Clinic College of Medicine, MN, USA; Albano and University la Sapienza, Italy;
University of Leipzig, Germany; Endocrinology Associates, AR, USA; Arcispedale Santa Maria Nuova,
Italy; Odense University Hospital, Denmark
GUIDE/NEWGUIDE
Dazzi et al., 1997 GUIDE Authoring and Execution Consorzio di Bioingegneria e Informatica Medica, Pavia, Italy
Quaglini et al., 1998 GUIDE Authoring and Execution
Consorzio di Bioingegneria e Informatica Medica, Pavia; University of Pavia; Istituto Nazionale per lo
Studio e la Cura dei Tumori, Milano, Italy Quaglini et al., 1999 GUIDE Authoring and Execution University of Pavia, Italy Quaglini et al., 2000 GUIDE Authoring and Execution University of Pavia, Italy
Quaglini et al., 2000 GUIDE Authoring and Execution University of Pavia; Consorzio di Bioingegneria e Informatica Medica, Pavia, Italy Peleg et al., 2003 GUIDE Authoring (case study) University of Pavia, Italy (GUIDE part) Ciccarese et al., 2003 NEWGUIDE Authoring and Execution
University of Pavia, Italy; Consorzio di Bioingegneria e Informatica Medica, Pavia, Italy; Ness-ISI Ltd., Beer
Sheva, Israel Ciccarese et al., 2004 NEWGUIDE Authoring and Execution University of Pavia, Italy Ciccarese et al., 2005 GUIDE/NEWGUIDE Authoring and Execution University of Pavia, Italy
GEM
Shiffman et al., 2001 GEM Cutter Mark-up tool Yale School of Medicine, CT, USA Shiffman et al., 2004 GEM Cutter Mark-up tool Yale School of Medicine, CT, USA Shiffman et al., 2004 GEM Cutter Mark-up tool Yale School of Medicine, CT; Yale New Haven
Hospital, CT, USA. Koch, et al., 2010 GEM Cutter Mark-up tool Mayo Clinic, MN, USA
DeGeL
Shahar, 2002 Uruz Mark-up tool Ben-Gurion University of the Negev, Israel Shahar et al., 2003 Uruz Mark-up tool Ben-Gurion University of the Negev, Israel Shahar et al., 2003 Uruz Mark-up tool Ben-Gurion University of the Negev, Israel Shahar et al., 2004 Uruz/Spock Mark-up tool/Execution
Engine Ben-Gurion University of the Negev, Israel Shalom and Shahar, 2005 Gesher Mark-up tool Ben-Gurion University of the Negev, Israel Young and Shahar, 2005 Spock Execution engine Ben-Gurion University of the Negev, Israel Young and Shahar, 2005 Spock Execution engine Ben-Gurion University of the Negev, Israel Shahar, 2006 Uruz/Spock Mark-up/Execution Engine Ben-Gurion University of the Negev, Israel Young et al., 2007 Spock Execution engine
Ben-Gurion University of the Negev, Israel; Wolfson Medical Center, Israel; Stanford University School of Medicine, CA, USA; Palo Alto Health Care System,
CA, USA; GRECC, CA, USA Hatsek et al., 2008 Gesher/Spock Mark-up/Execution Engine Ben-Gurion University of the Negev, Israel Hatsek et al., 2008 Gesher/Spock Mark-up/Execution Engine Ben-Gurion University of the Negev, Israel Hatsek et al., 2010 Gesher Mark-up tool Ben-Gurion University of the Negev, Israel
Topaz et al., 2013 Gesher Mark-up tool
University of Pennsylvania School of Nursing, PA, University of Texas School of Nursing, TX, & University of Minnesota School of Nursing, MN, USA;
Ben Gurion University of the Negev, Israel Shalom et al., 2016 Picard GBDSS Ben Gurion University of the Negev, Israel; Soroka
Medical Center, Israel
SAGE
Campbell et al., 2003 SAGE (Protégé based) Authoring and Execution
University of Nebraska Medical Center, NE; Stanford University Medical Center, CA; IDX Systems Corporation, WA; Intermountain Health Care, UT;
Mayo Medical School, MN; USA Ram et al., 2004 SAGE (Protégé based) Authoring and Execution IDX Systems Corporation, WA, USA Shankar et al., 2004 KWIZ Authoring University of Newcastle, UK; Stanford University School of Medicine, CA, US
Berg et al., 2004 SAGE Desktop Authoring IDX Systems Corporation, WA, USA Tu et al., 2007 SAGE (Protégé based) Authoring and Execution
Stanford University School of Medicine, CA; University of Nebraska Medical Center, NE; GE Healthcare Integrated IT Solutions, WA; Mayo Clinic, MN; Apelon
24 |
P a g e
Inc., CT; RemedyMD Inc., UT; Hospira, Inc., IL; Kea Analytics, WA; USA
Kim et al., 2008 SAGE (Protégé based) Authoring Kwandong University; InHa University; Seoul National University; South Korea Lee et al., 2010 SAGE + (uEngine +
BRAIN)
Authoring, inference + rule engines
Ajou University; Kwandong University; Inha University; Seoul National University, South Korea Yin et al., 2013 SAGE module GBDSS Zhejiang University, China
Liu et al., 2014 SAGE (Protégé based) Authoring Fourth Military Medical University; China Three Gorges University; China
Kim et al., 2015 SAGE + uBRAIN Authoring, inference + rule engines
Eulji University; Seoul National University; InHa University; Kwandong University: South Korea;
University of Utah, UT, USA
HELEN
Skonetzki et al., 2004 HELEN Authoring and Execution Heidelberg University Medical Center, Germany
SEMANTIC WEB & INFERENCE ENGINES
Abidi et al., 2007 GEM/CPG-EX Authoring and Execution Dalhousie University, Canada Hussain and Abidi, 2008 CPG-EX Authoring and Execution Dalhousie University, Canada Din et al., 2010 Protégé + Jena Authoring and Execution Dalhousie University, Canada Chen et al., 2011 Protégé + Jess Authoring and Execution Taipei Medical University; National Chengchi
University; Taiwan
Papageorgiou et al., 2012 FCM-uUTI DSS GBDSS Technological Educational Institute of Lamia, Greece; Agfa HealthCare NV, Belgium
Papageorgiou, 2012 FCM-uUTI DSS GBDSS Tech. Educational Institute of Lamia, Greece Linan et al., 2015 HeD editor Authoring Arizona State University, AZ; Mayo Clinic, MN; USA Jafarpour et al., 2016 OWL 1 DL, OWL 2 DL,
OWL 2 DL + SWRL Execution engine Dalhousie University, Canada
OTHER
SIEGFRIED
Lobach et al., 1998 SIEGFRIED GBDSS Duke University Medical Center, NC, USA Porcelli and Lobach, 1999 SIEGFRIED GBDSS Wake Forest University School of Medicine, NC, USA;
Duke University Medical Center, NC, USA
HGML
Hagerty et al., 2000 HGML Markup tool Mark-up tool Rutgers University, NJ, USA
STEPPER
Svatek and Ruzicka, 2002 Stepper Mark-up tool Cardio University of Economics, Czech Republic Svatek and Ruzicka, 2003 Stepper Mark-up tool Cardio University of Economics, Czech Republic Ruzicka and Svatek, 2004 Stepper Mark-up tool Cardio University of Economics, Czech Republic
MEDICAL MARKUP TOOL
Kumar et al., 2003 MTM Tool Mark-up tool University of Pavia, Italy