A Meta-model for the Guideline Definition Language
Reyes Grangel
1 a
, Cristina Campos Sancho
1 b
, Bego
˜
na Mart
´
ınez-Salvador
2 c
and Mar Marcos
2 d
1
Departament de Llenguatges i Sistemes Inform
`
atics, Universitat Jaume I, Castell
´
o, Spain
2
Departament d’Enginyeria i Ci
`
encia dels Computadors, Universitat Jaume I, Castell
´
o, Spain
Keywords:
Computer-Interpretable Guidelines, openEHR, GDL, Meta-model, Case Study.
Abstract:
Computer-Interpretable Guidelines (CIGs) are a key issue to implement decision support systems that could
help clinical practice. To represent these guidelines numerous modeling languages, such as PROforma, Asbru,
or GLIF, have been defined and they are currently used in different contexts with more or less acceptance. The
Guideline Definition Language (GDL) is a rule-based modeling language for CIGs proposed recently by the
openEHR Foundation. This language might have a wider acceptance because of the fact that it is defined
as an open standard and, as such, open detailed specifications are provided. In this paper, we focus on the
most recent specification of GDL, GDL2. Our objective is to gain insight and knowledge about this language
and its specifications for software engineering application purposes, such as Model-Driven solutions. In this
context, we present a proposal for a GDL meta-model as an attempt to formalise the GDL2 specification in a
UML meta-model. Additionally, in order to validate this proposal, we present a sample GDL model, based on
a clinical guideline for the diagnosis of heart failure, developed using the GDL meta-model implemented in
Eclipse.
1 INTRODUCTION
According to the most recent definition (Graham
et al., 2011), clinical practice guidelines are defined
as ‘statements that include recommendations intended
to optimize patient care that are informed by a sys-
tematic review of evidence and an assessment of the
benefits and harms of alternative care options’. The
definition of clinical practice guidelines (henceforth,
clinical guidelines) is a complex task that requires the
participation of expert groups and many evaluation
and verification activities. However, clinical guide-
lines have become a powerful resource for practition-
ers during decision-making processes in daily care
practice.
Aiming to obtain guidelines that were computer-
interpretable, and therefore provide the basis for clin-
ical decision support systems, several representation
languages and notations have been proposed in the
field of Artificial Intelligence and Medical Informat-
ics. The representation of the clinical guidelines us-
ing these modeling languages and notations are often
a
https://orcid.org/0000-0002-4049-3888
b
https://orcid.org/0000-0003-1140-1603
c
https://orcid.org/0000-0001-5959-9415
d
https://orcid.org/0000-0001-9672-4190
called Computer-Interpretable Guidelines (CIGs).
However, one critical issue that arises in practice
is that none of these numerous existing languages
(de Clercq et al., 2004), (Peleg et al., 2003), Arden
Syntax, PROforma, Asbru, EON, Prodigy, GLIF, and
GUIDE, has become a de facto standard to be widely
used by the scientific and medical community.
In this context, the openEHR Foundation
(openEHR Foundation, 2021) proposes a set of open
standards to represent this kind of guidelines aiming
mainly to provide an open framework to manage
Electronic Health Records (EHR) in an interoperable
way. Two of the included specifications are the
Guideline Definition Language (GDL), that is a
rule-based language to represent clinical guidelines,
and the Task Planning (TP) language jointly with
its graphical version the Task Planning Visual Mod-
elling Language (TP-VML) to support clinical task
planning processes.
Even though openEHR Foundation provides a de-
tailed specification of GDL2, this specification can-
not be considered a formal meta-model to be used for
software engineering purposes. The research work
presented in this paper includes a first proposal of a
meta-model for GDL, that is completed and validated
by means of a case study developed for the clinical
Grangel, R., Campos Sancho, C., Martínez-Salvador, B. and Marcos, M.
A Meta-model for the Guideline Definition Language.
DOI: 10.5220/0010543400590068
In Proceedings of the 16th International Conference on Software Technologies (ICSOFT 2021), pages 59-68
ISBN: 978-989-758-523-4
Copyright
c
2021 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
59
guideline dealing with the diagnosis and management
of heart failure (The Task Force for the diagnosis and
treatment of acute and chronic heart failure of the Eu-
ropean Society of Cardiology (ESC), 2016).
The paper is organised as follows. Section 2 out-
lines the standards related to healthcare used in this
paper, and the background in the context of meta-
modeling. Section 3 shows the methodology fol-
lowed in order to obtain the GDL meta-model and the
meta-model proposal itself. Finally, the case study
is described in Section 4, and Section 5 presents the
conclusions together with some prospects for future
work.
2 HEALTHCARE AND
SOFTWARE ENGINEERING
BACKGROUND
This section analyses the main concepts used in the
research work linking healthcare and software engi-
neering context.
2.1 openEHR Standards
The openEHR Foundation is one of the most note-
worthy international standards development organisa-
tions inside the healthcare community. The mission
of this international non-profit organisation is to en-
able semantic interoperability of health information,
within and between EHR systems (Health Informa-
tion and Quality Authority (HIQA), 2013). One of
openEHR Foundation main achievements is to have
provided widely accepted open specifications in this
field.
openEHR Specifications is one of the openEHR
Foundation initiatives that provides technology for e-
health, which includes open specifications, clinical
models, and software for researchers and develop-
ers, so that they can create interoperable solutions for
healthcare (openEHR Foundation, 2021). These spec-
ifications and solutions are based on the requirements
captured over many years in projects such as the EU
FP3 Good European Health Record (GEHR) project
(1992-1995), with the aim of developing systems that
would be interoperable.
Figure 1 shows the openEHR architecture includ-
ing the abstract specifications, also known as the
‘Platform Independent Model’ (PIM). The specifica-
tions relevant to CIG modeling are the Archetypes
shown as a Formalisms component, the openEHR
Terminology included as a Content component, and
‘Process & CDS’ (CDS for Clinical Decision Sup-
port). This last component includes the two main
specifications regarding the guideline modeling, the
GDL Language, for representing the decision rules of
guidelines and the Task Planning Language, for rep-
resenting the clinical processes of guidelines.
Figure 1: The openEHR Specification project (openEHR
Foundation, 2021).
The knowledge contained in clinical guidelines is
not only related to the decision-making processes that
can be represented by rules, but also to patient data or
to the tasks that need to be planned in the course of pa-
tient management. According to (Peleg et al., 2003),
two main categories of knowledge could be identified
in CPGs: structuring in plans of decisions and actions,
and linking to patient data and medical concepts.
Modeling plans using GDL, i.e., modeling the
workflow of clinical tasks to be performed, is a hard
issue. According to (Marcos et al., 2019), it re-
quires the use of ad hoc variables and a great number
of rules in order to represent correctly the execution
flow of the actions. The openEHR architecture in-
cludes within the Process and Clinical Decision Sup-
port module an additional specification for Task Plan-
ning (TP), as well as its graphical version, Task Plan-
ning Visual Modelling Language (TP-VML), which
are useful to support flow control execution.
2.2 GDL
GDL is one of the specifications provided by
openEHR community to represent part of the knowl-
edge, specifically the decision logic, included in clin-
ical guidelines (openEHR Foundation, 2019). The
main advantage of this language over other for-
malisms developed in this context, such as PROforma
(Sutton and Fox, 2003) or Asbru (Seyfang et al.,
2002), is that it is based on an open standard. More-
over, it incorporates openEHR archetypes (openEHR
Foundation, 2007) for the description of guideline
data.
GDL is a formal language whose objective is to
express decision support logic as a set of produc-
tion rules. GDL rules contain ‘when-then’ statements,
ICSOFT 2021 - 16th International Conference on Software Technologies
60
which can be combined together as building blocks
to support single decision making as well as more
complex, chained, decision making processes. In this
sense, GDL rules are suitable to be used in the task
of modeling CIGs that can be part of decision support
systems for healthcare (openEHR Foundation, 2021).
This language uses archetypes to model the clini-
cal data collected and required in clinical guidelines.
There exist some works that show how to use GDL
jointly with archetypes in order to represent clinical
guidelines in a rule-based manner (Anani et al., 2014;
Lin et al., 2016).
GDL2 is the evolution of GDL version 1
1
. A dis-
tinctive characteristic of GDL2 with respect to GDL
version 1 is that it provides a specification structured
in packages. Although this structure cannot be used
as a meta-model, it is a good starting point to de-
velop it. The packages defined in the GDL2 specifica-
tions (openEHR Foundation, 2021) are the following:
Guideline Package, which includes the classes
needed to identify the archetypes used in the
guideline and the defined rules.
Expression Package, which incorporates the
needed classes to make up the expressions (asser-
tions and assignments) included in the rules, pre-
conditions and default actions.
Terminology Package, which incorporates the
needed classes to define the terms and the term
bindings.
GDL2 basically provides preconditions, default ac-
tions and rules to execute the logic of the guideline
decisions, and openEHR archetypes used to interop-
erate with the EHR.
A GDL model includes a description, a definition
and a terminology section. The description section
contains the identification of the guideline: name, au-
thors, version, keywords, etc.
The definition section includes the archetypes in-
stantiation, the preconditions, the default actions and
a list of rules to be executed. Besides, it can define
different output formats that are identified in an out-
put template. An archetype describes in detail the
structure and content of clinical concepts. In a GDL
model archetypes can be instantiated as input or out-
put archetypes to deal with the input and output data
respectively. Preconditions are defined assertions that
have to be met before the rules of the guideline can
be executed. Default actions define assignments to be
always executed. A rule is defined by means of an
assertion (when) and one or more assignments (then),
1
In this paper, the terms ‘GDL’ and ‘GDL2’ are equally
used to refer to the version 2 of GDL, while ‘GDL version
1’ is used to identify the version 1 of GDL.
i.e., (if-then) statements. Rules are executed follow-
ing the priority provided by their order. Assertions
and assignments are logical expressions composed
by items and operators. Items can be in turn logi-
cal expressions, data elements instantiated from the
archetypes or constant values. Finally, the terminol-
ogy section defines how the terms used in the guide-
line are linked to user interface labels and to the term
description in natural language. Moreover, it can in-
clude the local terms bound to external concepts by
means of external references named term bindings.
Figure 4 shows an example of the rules defined
for a GDL model implemented using the GDL2 Edi-
tor Primer
2
. Figure 5 shows the rule Step 1 HF or
other diagnoses decision - suspected. First
the rule sets up the the conditions composed by log-
ical expressions that assess values of the instantiated
elements from archetypes, and then it includes the ac-
tions to be executed if the previous nested conditions
are true.
2.3 Use of Meta-models in Software
Engineering
The model and meta-model concepts are usually con-
fused in the context of Software Engineering, often
due to the numerous definitions that can be found in
the literature. Therefore, and according to (Do et al.,
2012), it is needed to specify the model and meta-
model concepts for a better comprehension.
A model is an abstract representation of the real
world or more specifically of a system. Therefore,
it must necessarily be a simplification of that reality
developed with a specific goal and purpose (Favre,
2004).
On the other hand, a meta-model is a model it-
self, but moreover, it is the explicit specification of
an abstraction. It needs to set up a list of constructs,
concepts and relationships among then, in order to de-
velop the abstraction, that is to say the corresponding
models (B
´
ezivin and Gerb
´
e, 2001).
Therefore, the meta-model represents the infor-
mation of the model, which represents the real world
that we want to abstract. In fact, the representation
by means of the concepts and relationships included
in the meta-model becomes then our understanding of
the real world (Do et al., 2012).
According to (Garc
´
ıa-Magari
˜
no et al., 2010)
meta-models need to formally specify modeling lan-
guages, providing a precise definition of the primi-
tives, constraints and rules required for creating mod-
2
https://gdl-lang.org/the-project/guides-tutorials/gdl2/
gdl2-editor-primer/
A Meta-model for the Guideline Definition Language
61
els given a specific problem or with one particular
goal.
In the software engineering domain, the meta-
modeling approach can be used to define domain-
specific modeling languages for one specific context
(Cuadrado and Molina, 2009; Viana et al., 2013), but
also to provide an improved specification of some
modeling languages whose syntax and semantics are
expressed in other kind of formalisms (Karagiannis
and K
¨
uhn, 2002). The final aim of this enhanced spec-
ification would be to provide more insight and knowl-
edge about the modelling language, and also to ap-
ply other approaches such as Model-Driven solutions
(B
´
ezivin and Gerb
´
e, 2001).
There exist different meta-modeling approaches
according to (Karagiannis and K
¨
uhn, 2002). In this
research we follow the OMG proposal Meta Object
Facility (OMG, 2021b), which provides four layers of
abstraction for meta-modeling: the Meta-Metamodel
layer (M3), with meta-modeling languages like MOF
and Ecore; the Metamodel layer (M2), where the
meta-models of languages are defined using meta-
modeling languages from M3; the Model layer (M1),
with the specific models defined from those meta-
models in M2; and the User Objects layer (M0), with
the concrete objects of a system as instances of mod-
els in M1. There are several available alternatives as
meta-modelling languages, such as Ecore, MOF and
Graph, Object, Property, Relationship, and Role (GO-
PRR) (Garc
´
ıa-Magari
˜
no et al., 2010).
2.4 Related Work
There are no examples of the use of meta-modelling
in the context of openEHR specifications. However,
there are numerous research works trying to apply
the meta-modeling approach to the healthcare domain
with different purposes and results.
An example of using meta-modeling to avoid the
problem that supposes to use a HL7 proprietary model
language that is not supported by all Model-Driven
Engineering (MDE) tools is shown in (Mart
´
ınez-
Garc
´
ıa et al., 2015). The proposal uses HL7 in a MDE
approach considering the MIF (Model Interchange
Format) meta-model proposed by HL7 by making use
of a plug-in developed in the EA (Enterprise Archi-
tect) tool.
We can find a lot of experiences in the MDE con-
text applied to the healthcare domain in which the use
of a meta-modelling approach is followed. In this
sense, a formalization is presented in (Rabbi et al.,
2014) to coordinate the integration of different meta-
models proposed in order to capture the complexity
of healthcare systems with purposes of software de-
velopment. While this is a more theoretical work,
it shows how meta-models are useful when trying to
represent the complexity inherent to healthcare sys-
tems, and how then, they can be used as part of an
MDE approach to generate software.
Other kinds of research works are focused on inte-
grating different modeling proposals of specific areas
of healthcare software systems. For instance, a Dis-
aster Management Meta-model is proposed in (Oth-
man et al., 2014) using different models to unify ap-
proaches, which allows knowledge sharing and effi-
cient disaster activities management. Besides, this
work provides a wide description of the meta-model
development and validation process followed.
Unlike the previously described results, the meta-
model developed in this work is based on the defined
openEHR specifications and the proposal is validated
using a case study.
3 GDL META-MODEL
PROPOSAL
This section describes the steps followed in order to
obtain the GDL meta-model proposal, and its descrip-
tion.
3.1 Methodology
The methodology used to obtain the meta-model is
based on Action-Research approach (Baskerville and
Myers, 2004). The steps followed are: domain analy-
sis, interpretation, and meta-model validation.
Domain analysis has consisted in identifying the
concepts and relationships that are needed to define
a GDL meta-model. To identify these contents we
have reviewed and analyzed both the GDL2 specifi-
cation provided by openEHR (openEHR Foundation,
2021), and the last version of the open source mod-
elling tool, the GDL2 Editor Primer developed by
Cambio Healthcare Systems (Cambio Group, 2021).
This meta-model is focused on the representation of
rules and terms, and for the time being, the modeling
of archetypes has not been considered.
The interpretation of previous results has led us to
define the concepts that make up the meta-model. We
have applied different iterations in this phase until we
have obtained a meta-model that allows designing a
GDL model equivalent to the model created with the
GDL2 Editor Primer. The tasks performed in each
iteration are:
Creation of the GDL meta-model using the
ICSOFT 2021 - 16th International Conference on Software Technologies
62
Eclipse Papyrus tool
3
.
Generation of the editing modeling tool based on
this meta-model using the Enterprise Modeling
Framework (EMF) utilities
4
.
Development of a model for an excerpt of a case
study using the generated tool, and assessment
(expressiveness, completeness, traceability, etc.)
of meta-model constructs.
Comparison and analysis of the developed model
with respect to the one developed using the GDL2
Editor Primer. Conclusions drawn from the result
of each iteration have served to improve the meta-
model versions.
The 2016 ESC Guidelines for the diagnosis and treat-
ment of acute and chronic heart failure (The Task
Force for the diagnosis and treatment of acute and
chronic heart failure of the European Society of Car-
diology (ESC), 2016) has been used both as a case
study to review the different iterations of the meta-
model and to validate the final GDL meta-model. In
order to achieve these two goals we have modeled an
excerpt of this guideline using the modeling tool gen-
erated from the designed Papyrus meta-model, and
also using the GDL2 Editor Primer.
3.2 Meta-model Description
Technically, the definition of a meta-model with the
aim of generating a modeling editor tool by means
EMF has specific requirements apart from the nor-
mal classes and relationships that make up the meta-
model. The meta-model must adopt certain rules so
that the generated modelling tool works correctly to
generate the corresponding model. First, it is needed
to define as a directed composition the relationship
between two classes to generate an element (target
of composite) from the other (source of composite)
into the model. Besides, the relationship between
two classes of the meta-model must be a normal di-
rected association in order to establish a relationship
between two elements in the model.
In the GDL meta-model proposal (see Figure 2)
the classes representing the three sections that com-
pose a GDL model (description, definition and ter-
minology) are clearly shown in the upper part of the
Figure.
Regarding the description section the classes de-
fined are:
Description: this class represents the identifying
information of the Guideline. Each Guideline has
3
https://www.eclipse.org/papyrus/
4
https://www.eclipse.org/modeling/emf/
a single description section including relevant in-
formation about the owner, author and contribu-
tors, descriptive data, version and keywords. Key-
word, Contributor and Reference are classes as
well, since they can be multi-valued concepts.
Keyword: it represents each concept that can be
defined as a tag within a Guideline.
Contributor: it represents each person who has
participated in the Guideline development.
Reference: it represents multiple links to informa-
tion sources regarding the Guideline.
Regarding the definition section the main class is Def-
inition, that represents the contents needed to design
a GDL model. These contents are DataBinding that
sets up the archetype list, a list of rules, a list of pre-
conditions, and a list of default actions. Moreover, a
definition section can define different output formats
that are identified in an output template.
According to the DataBinding concepts the
classes defined are:
DataBinding: this class defines the set of spe-
cific archetypes (ArchetypeInstantiation) used in
the model.
ArchetypeInstantiation: it represents each single
archetype in a GDL guideline.
ElementInstantiation: it defines each single in-
stantiated element from an instantiated archetype.
Regarding the rules, the classes included are:
RuleList: this class represents the set of rules de-
fined for the execution of the Guideline. A Guide-
line has a single RuleList.
Rule: it represents each pair of conditions and ac-
tions defined for the execution of the guideline.
Each rule has an identification and a priority that
sets the execution order. A rule is composed by
one or more expressions, as the condition or the
when part, that must be met to fire the rule, and a
set of one or more actions, as the then part of the
rule that are executed if the when part is met.
RuleCondition: it represents the Boolean expres-
sions that must be assessed in a rule.
RuleAction: it represents the Boolean expressions
that assign a value or an expression of any type to
a variable (ElementInstatiation).
The other classes related to the GDL definition sec-
tion are:
PreconditionList: it represents the set of logical
conditions that must be met before the execution
of the Guideline.
A Meta-model for the Guideline Definition Language
63
Figure 2: UML representation of the GDL meta-model proposal.
ICSOFT 2021 - 16th International Conference on Software Technologies
64
Precondition: it represents the preconditions that
apply to the guideline as a whole.
DefaultActionsList: it represents actions defined
to be executed without any previous condition.
Template: this class represents each defined struc-
tured format for the output data reporting the re-
sult of the guideline execution.
Conditions, pre-condition and actions are defined us-
ing expressions represented by the following classes:
Expression: it represents the logical statements to
be assessed or assigned. Expressions are defined
for the rules (conditions and actions), precondi-
tions and default actions.
ExpressionItem: it represents the elements used in
the expressions. An expression has three kind of
elements: constants, operations and variables.
ElementInstantiationExpression: this class repre-
sents the use of an element (ElementInstantiation)
in an expression.
Regarding the terminology section the classes are:
Terminology: this class represents the section
which includes the definition of the semantic
terms (rules, data, etc.) used in the guideline.
TermList: it represents the set of terms that are
used in the guideline.
Term: this class represents a concrete concept and
its tag that can be used inside the guideline.
TermBinding: it represents links to external ter-
minology.
4 CASE STUDY FOR
META-MODEL VALIDATION
The case study selected to validate the GDL meta-
model is the 2016 ESC Guidelines for the diagno-
sis and treatment of acute and chronic heart failure
(henceforth, ESC-HF guideline) (The Task Force for
the diagnosis and treatment of acute and chronic heart
failure of the European Society of Cardiology (ESC),
2016).
As mentioned before, we have considered a small
part of this ESC-HF guideline focused on the algo-
rithm for the diagnosis of the heart failure in the non-
acute setting (see Figure 3). This excerpt of the ESC-
HF guideline had been already modelled in GDL ver-
sion 1 in a previous research work (Marcos et al.,
2019). In this work we have used the new version
of the GDL2 modeling tool to review and update this
algorithm, which is presented in Figure 4 and 5.
Figure 3: Diagnostic algorithm included in ESC-HF guide-
line (The Task Force for the diagnosis and treatment of
acute and chronic heart failure of the European Society of
Cardiology (ESC), 2016).
Figure 4: Rules of the GDL model for the case study.
The set of rules defined for this excerpt is shown
in Figure 4. The rule order sets up the priority in
the rule execution. From the list of rules, the rule
selected in red, Step 1 HF or other diagnoses
decision - suspected, is shown in detail with its
A Meta-model for the Guideline Definition Language
65
assertions and actions in Figure 5. This rule models
the assessment of HF probability shown in Figure 3.
Figure 5: Detail of the rule Step 1 HF or other
diagnoses decision - suspected.
As it can be seen in Figure 5, the rule has a com-
pound assertion in the Rule Condition panel, and two
rule actions in the Rule Actions panel. The asser-
tion evaluates whether ‘HF or other diagnoses deci-
sion CG’ exists, i.e., has a defined value, and if one
of the conditions in the ‘or’ expression (‘Clinical his-
tory clues = true’, or ‘ECG results normal CG = false’
or ‘Physical examination clues = true’) is true. Then,
if the condition is accomplished the two actions that
assign specific values to the elements ‘HF or other di-
agnoses decision’ and ‘HF diagnosis status’ are exe-
cuted.
Once the GDL meta-model was defined and using
the Eclipse EMF plug-in, an editing tool based on the
meta-model was obtained. Figure 6 shows the same
excerpt of the guideline shown in Figure 3 but this
time modeled with the editing tool based on the GDL
meta-model. The left-hand side of Figure 6 shows the
overview tree of the implemented guideline. The defi-
nition part includes the data binding using archetypes
instantiation and the set of rules defined for model-
ing the diagnostic algorithm. It can be observed at a
glance that this set of rules is the same that the one
shown if Figure 4 implemented with the GDL2 Edi-
tor Primer. The right-hand side of the Figure 6 shows
with more detail the same rule displayed in Figure 5.
On the top, in first place, the tree of the rule condition
(when part) is shown. This condition comprises four
expressions joined with logical or operators. Here-
after, the details of the two rule actions are shown
(then part). Each rule action is an expression which
is composed in turn of two expression items. Finally,
at the bottom of the Figure 6, the properties window
of the rule is shown. One of the elements of the prop-
erties is the term for labelling the rule.
We can conclude comparing both models, pre-
sented in Figures 4 and 5, and in Figure 6, that
the modeling tool generated from the proposed GDL
meta-model allows to properly define the ESC-HF
guideline. Thus, the meta-model comprises and en-
ables the definition of all the elements needed in a
guideline according to the GDL specifications.
5 CONCLUSION
GDL is a rule-based language suitable to represent
part of the knowledge included in clinical guidelines
based on open specifications provided by openEHR
Foundation.
In this paper, we present a proposal for a GDL
meta-model that could be used jointly with these open
specifications by the scientific and medical commu-
nity. Furthermore, the meta-model is validated by
means of a case study based on the diagnosis of the
heart failure in the non-acute setting.
Taking into account that the new version of GDL,
GDL2, has a specification but not a formal meta-
model, the main contribution of this research work is
to provide a meta-model useful in the software engi-
neering context. For instance, this meta-model can
be used to solve problems related to the healthcare
domain, as for example when it comes to applying a
MDE approach.
Moreover, the meta-model contributes to a deeper
comprehension of GDL specifications and this fact
could increase the applied use of this language by
health care practitioners.
The limitations of the work are related to the val-
idation process. More experiences with different and
wider case studies would be carried out in order to
improve and refine the current version of the GDL
meta-model. Besides, the meta-model includes the
needed classes to instantiate archetypes although their
detailed modeling has not been considered because it
is out of scope of this project. However, it is a ques-
tion that will be addressed in future research work.
Finally, the future research work planned is to use
this GDL meta-model in a MDE project to transform
BPMN Models (OMG, 2021a) into GDL models, in
a similar way that the work performed in (Mart
´
ınez-
Salvador et al., 2015), but using GDL instead of PRO-
forma and following a Model-Driven approach.
ICSOFT 2021 - 16th International Conference on Software Technologies
66
Figure 6: GDL model of the case study defined using the editing tool based on the GDL meta-model.
ACKNOWLEDGEMENTS
This work has been partially supported by the Spanish
Ministry of Economy and Competitiveness through
the grant TIN2014-53749-C2-1-R.
REFERENCES
Anani, N., Chen, R., Moreira, T. P., and Koch, S. (2014).
Retrospective checking of compliance with practice
guidelines for acute stroke care: a novel experi-
ment using openEHR’s Guideline Definition Lan-
guage. BMC medical informatics and decision mak-
ing, 14(1):1 – 18.
Baskerville, R. and Myers, M. D. (2004). Special Issue on
Action Research in Information Systems: Making IS
Research Relevant to Practice: Foreword. MIS Quar-
terly, 28(3):329 – 335.
B
´
ezivin, J. and Gerb
´
e, O. (2001). Towards a precise defini-
tion of the OMG/MDA framework. In Proceedings of
16th Annual International Conference on Automated
Software Engineering, 2001 (ASE 2001), pages 273
280. IEEE.
Cambio Group (2021). GDL2 Editor Primer. In https://
www.cambiogroup.com/. Accessed: 2021-02-26.
Cuadrado, J. S. and Molina, J. (2009). A Model-Based
Approach to Families of Embedded Domain-Specific
Languages. Software Engineering, IEEE Transactions
on, 35:825 – 840.
de Clercq, P. A., Blom, J. A., Korsten, H. H. M., and Has-
man, A. (2004). Approaches for creating computer-
interpretable guidelines that facilitate decision sup-
port. Artificial Intelligence in Medicine, 31(1):1 – 27.
Do, Q., Cook, S. C., Campbell, P., Scott, W., Robinson, K.,
Power, W., and Tramoundanis, D. (2012). Require-
ments for a Metamodel to Facilitate Knowledge Shar-
ing between Project Stakeholders. In Dagli, C. H.,
editor, Proceedings of the Conference on Systems En-
gineering Research, CSER 2012, St. Louis, MO, USA,
March 19-22, 2012, volume 8 of Procedia Computer
Science, pages 285 – 292. Elsevier.
Favre, J.-M. (2004). Towards a basic theory to model model
driven engineering. 3rd Workshop in Software Model
Engineering, WiSME.
Garc
´
ıa-Magari
˜
no, I., Fuentes-Fern
´
andez, R., and G
´
omez-
Sanz, J. J. (2010). A framework for the definition
of metamodels for Computer-Aided Software Engi-
neering tools. Information and Software Technology,
52(4):422 – 435.
Graham, R., Mancher, M., Wolman, D. M., Greenfiled, S.,
and Steinberg, E. (2011). Institute of Medicine: Clin-
ical Practice Guidelines We Can Trust. The National
Academies Press.
Health Information and Quality Authority (HIQA) (2013).
Overview of Healthcare Interoperability Standards.
Karagiannis, D. and K
¨
uhn, H. (2002). Metamodelling Plat-
forms. In Bauknecht, K., Tjoa, A. M., and Quirch-
mayr, G., editors, E-Commerce and Web Technolo-
gies, pages 182 – 196. Springer Berlin Heidelberg.
Lin, C.-H., Lo, Y.-C., Hung, P.-Y., and Liou, D.-M. (2016).
A Meta-model for the Guideline Definition Language
67
Building Chronic Kidney Disease Clinical Practice
Guidelines Using the openEHR Guideline Definition
Language. Methods of information in medicine.
Marcos, M., Campos, C., and Mart
´
ınez-Salvador, B.
(2019). A Practical Exercise on Re-engineering Clini-
cal Guideline Models Using Different Representation
Languages. In Marcos, M., Juarez, J. M., Lenz, R.,
Nalepa, G. J., Nowaczyk, S., Peleg, M., Stefanowski,
J., and Stiglic, G., editors, Artificial Intelligence in
Medicine: Knowledge Representation and Transpar-
ent and Explainable Systems - AIME 2019 Interna-
tional Workshops, KR4HC/ProHealth and TEAAM,
Poznan, Poland, June 26-29, 2019, Revised Selected
Papers, volume 11979 of Lecture Notes in Computer
Science, pages 3 – 16. Springer.
Mart
´
ınez-Garc
´
ıa, A., Garc
´
ıa-Garc
´
ıa, J. A., Escalona, M. J.,
and Parra-Calder
´
on, C. L. (2015). Working with the
HL7 metamodel in a Model Driven Engineering con-
text. Journal of Biomedical Informatics, 57:415 424.
Mart
´
ınez-Salvador, B., Marcos, M., and Riano, D. (2015).
An Algorithm for Guideline Transformation: from
BPMN to SDA. The 5th International Conference on
Current and Future Trends of Information and Com-
munication Technologies in Healthcare (ICTH 2015),
63:244 – 251.
OMG (2021a). Business Process Model and Notation. In
http://www.bpmn.org/. Accessed: 2021-02-26.
OMG (2021b). The MetaObject Facility Specification. In
https://www.omg.org/mof/. Accessed: 2021-02-26.
openEHR Foundation (2007). Archetype Definitions and
Principles. In https://specifications.openehr.org/
releases/1.0.2/architecture/am/archetype principles.
pdf. Accessed: 2021-02-26.
openEHR Foundation (2019). Guideline Definition Lan-
guage v2 (GDL2). In https://specifications-test.
openehr.org/releases/CDS/latest/GDL2.html. Ac-
cessed: 2021-02-26.
openEHR Foundation (2021). Open industry specifications,
models and software for e-health (openEHR). In https:
//www.openehr.org/. Accessed: 2021-02-26.
Othman, S. H., Beydoun, G., and Sugumaran, V. (2014).
Development and validation of a Disaster Manage-
ment Metamodel (DMM). Inf. Process. Manag.,
50(2):235 – 271.
Peleg, M., Tu, S., Bury, J., Ciccarese, P., Fox, J., Greenes,
R. A., Miksch, S., Quaglini, S., Seyfang, A., Short-
liffe, E. H., Stefanelli, M., and et al. (2003). Com-
paring Computer-Interpretable Guideline Models: A
Case-Study Approach. Journal of the American Med-
ical Informatics Association (JAMIA), 10:52 – 68.
Rabbi, F., Lamo, Y., and MacCaull, W. (2014). Co-
ordination of Multiple Metamodels, with Application
to Healthcare Systems. In The 5th International Con-
ference on Emerging Ubiquitous Systems and Perva-
sive Networks (EUSPN-2014)/ The 4th International
Conference on Current and Future Trends of Informa-
tion and Communication Technologies in Healthcare
(ICTH 2014)/ Affiliated Workshops, volume 37 of Pro-
cedia Computer Science, pages 473 – 480. Elsevier.
Seyfang, A., Kosara, R., and Miksch, S. (2002). Asbru’s
Reference Manual, Asbru Version 7.3.
Sutton, D. R. and Fox, J. (2003). The Syntax and Seman-
tics of the PROforma Guideline Modeling Language.
Journal of the American Medical Informatics Associ-
ation (JAMIA), 10(5):433 – 443.
The Task Force for the diagnosis and treatment of acute
and chronic heart failure of the European Society of
Cardiology (ESC) (2016). 2016 ESC Guidelines for
the diagnosis and treatment of acute and chronic heart
failure.
Viana, M. C., Penteado, R. A. D., and do Prado, A. F.
(2013). Domain-Specific Modeling Languages to im-
prove framework instantiation. Journal of Systems
and Software, 86(12):3123 – 3139.
ICSOFT 2021 - 16th International Conference on Software Technologies
68