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

Departament de Llenguatges i Sistemes Inform

atics, Universitat Jaume I, Castell

o, Spain

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

https://orcid.org/0000-0002-4049-3888

https://orcid.org/0000-0003-1140-1603

https://orcid.org/0000-0001-5959-9415

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

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

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

. 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),

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

. 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-

https://gdl-lang.org/the-project/guides-tutorials/gdl2/

gdl2-editor-primer/

A Meta-model for the Guideline Deﬁnition Language

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

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

Eclipse Papyrus tool

• Generation of the editing modeling tool based on

this meta-model using the Enterprise Modeling

Framework (EMF) utilities

• 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

https://www.eclipse.org/papyrus/

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

Figure 2: UML representation of the GDL meta-model proposal.

ICSOFT 2021 - 16th International Conference on Software Technologies

• 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

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

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.

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

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