Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways

Tri Huynh, Selin Erdem, Felix Eckert and Florian Matthes

School of Computation, Information and Technology, Technical University of Munich, Munich, Germany

Keywords:

Domain-Speciﬁc Language, Domain-Speciﬁc Modeling Language, Clinical Pathway Modeling, Adaptive

Case Management, Decision Support Tool.

Abstract:

e-Health systems leverage clinical pathways (CPs) models as standardized and optimized procedures to exe-

cute and manage medical treatments. To model CPs in decision support e-Health systems, our study devel-

ops Acadela, a low-tech-oriented, text-based Domain Speciﬁc Language (DSL) with visualization capability.

Acadela declares grammar to enforce textual syntax for modeling workﬂow, control ﬂow, responsibility, med-

ical data visualization, and communications with external systems. Furthermore, Acadela provides a model

visualization to preview the CP and assist communication between medical and technical experts. To explore

the DSL’s expressiveness and usability, we conducted two separate descriptive user studies with six medical

professionals and eight technical adepts. First, we modeled ﬁve CPs used by medical professionals in their

daily routines. Through semi-structured interviews, we collected feedback regarding the language’s expres-

siveness. Next, we invited the technical adepts to model a hypertension CP and debug a faulty model written in

Acadela. Overall, the medical experts consider the modeled CPs accurately reﬂect their treatment procedure;

and the technical adepts consider the language easy to use and applicable to model CPs. The results imply

the DSL’s potential to model CPs with various degrees of complexity in different medical ﬁelds while being

user-friendly to modelers.

1 INTRODUCTION

E-Health applications are systems that apply a com-

bination of electronic communication and informa-

tion technology to transmit, store and retrieve digi-

tal data electronically for educational, clinical, or ad-

ministrative purposes in the healthcare sector, both lo-

cally and remotely (Mitchell, 1999). Operating treat-

ment processes for patients in e-Health applications

requires modeling clinical pathways (CPs). In sum-

mary, CPs are patient-care management plans that de-

ﬁne care goals with the process and timing necessary

to achieve such goals with optimal efﬁciency (Panella

et al., 2003; Campbell et al., 1998; Every et al., 2000).

Applying CPs results in numerous beneﬁts, from im-

proving patient conditions (Panella et al., 2008), re-

ducing complications (Preston et al., 2013), mortality

(Hai et al., 2019), length of stay, and costs (Yang and

Su, 2014). State-of-the-art solutions propose graphi-

cal or textual Domain Speciﬁc Languages (DSLs) to

model CPs and leverage their beneﬁts in e-Health sys-

tems. However, each approach has distinct advan-

tages and limitations.

On the one hand, graphical DSLs enable model-

ers to build medical procedures by interacting with

visual elements. The advantages of this approach

are a user-friendly interface and learnable mechanism

(Hermans et al., 2009, p. 433), as the applications

display workﬂow elements following a logical and

hierarchical structure (Wienands and Golm, 2009, p.

458). Furthermore, visual artifacts foster communica-

tion among technical and domain experts (Heß et al.,

2015, p. 13), as they share a common understand-

ing of notations for CP elements. However, existing

workﬂow modeling languages, like BPMN, lack the

notation to express aspects besides workﬂow execu-

tion, thus demanding extra effort to develop custom

extensions for modeling and visualizing these aspects

(Heß et al., 2015, p. 4), (Braun et al., 2014, p. 10).

On the other hand, textual DSLs model CPs us-

ing a text-based interface and grammars (Cook et al.,

2007, pp. 15-17) or meta-models (Jouault et al.,

2006, p. 2). The textual model representation com-

bined with an IDE brings several beneﬁts. First, ex-

tending the model is convenient by textually deﬁn-

ing sub-DSLs of new domain concepts (Rieger et al.,

2018). Second, validating models is manageable as

textual DSLs only deﬁne constraints via grammar

(Baar, 2015). Third, an IDE offers convenient support

Huynh, T., Erdem, S., Eckert, F. and Matthes, F.

Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways.

DOI: 10.5220/0012076500003538

In Proceedings of the 18th International Conference on Software Technologies (ICSOFT 2023), pages 46-58

ISBN: 978-989-758-665-1; ISSN: 2184-2833

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

with error warning, syntax highlighting, and auto-

completion of CP elements and their values (Cook

et al., 2007, pp. 13-14), (Merkle, 2010). How-

ever, textual DSLs typically offer no CP visualiza-

tion, which is essential to present a comprehensive

overview using graphical notations (Frank, 2013, p.

134) familiar to both domain and technical experts,

thus fostering communication and workﬂow analysis

to increase the quality of the treatment.

Furthermore, from literature research, existing

DSLs for CPs do not support the modeling of coor-

dination with external systems or customizing the vi-

sual representation of medical data. External com-

munication is vital to obtain decision-support data

(e.g., adverse drug-drug interaction effects (Thakrar

et al., 2007)), orchestrate and integrate with existing

(legacy) systems (White, 2009, p. 11) (Kurz et al.,

2015, p. 4) (Iroju et al., 2013, pp. 265-268), as they

require necessary data (e.g., user information, medi-

cal status) from these systems to execute healthcare

services for each patient case. Meanwhile, customiz-

ing data representation provides ﬂexibility in display-

ing medical data (e.g., as a graph, image, or decorated

text (Michel, 2020, p. 36)) to express the patient’s

condition comprehensively and vividly. An under-

standable visualization is prominent in medical treat-

ment as they provide healthcare professionals with an

overarching and insightful overview of medical states

to ease decision-making, particularly in complicated

scenarios (Suganthi and Poongodi, 2021).

In addition to the technical functionality, a DSL

should be user-friendly to modelers. Therefore, we

also explore user-friendliness and learnability from

the modeler’s viewpoint in our study.

To address the missing essential CP concepts and

study the usability of our solution, we develop a

generic, textual DSL named Acadela for CP mod-

eling. Besides declaring essential CP elements,

Acadela can 1) deﬁne when and how to send HTTP

requests to external systems, 2) customize the graphi-

cal representation of medical data in the e-Health ap-

plication, and 3) visualize the modeled CP using ba-

sic shapes and color code. Our goal is to integrate

Acadela into e-Health applications without drastically

modifying the system while increasing the usability

and productivity of the CP modeling process.

The remainder of our paper consists of seven sec-

tions. First, we explained the research methodology

applied in our study. The next section presents the re-

lated work of existing DSLs that identify the required

concepts and features for modeling CPs discussed in

the fourth section. The ﬁfth section showcases con-

crete syntax to model the identiﬁed CP concepts in

our DSL. The sixth and seventh sections report our

descriptive user studies regarding the expressiveness

and usability of the DSL, respectively. The ﬁnal sec-

tion concludes our study and proposes future work.

2 METHODOLOGY

To envision the development and evaluation of our

DSL, we formulate three research questions focusing

on 1) the concrete syntax, 2) expressiveness, and 3)

usability as follows:

RQ1: How can a textual DSL model executable CPs

that supports external communication, customizable

graphical representation of medical data, and visual-

ization of CP models?

RQ2: Can the DSL model CPs from different medi-

cal ﬁelds with diverse complexity while being under-

standable to clinical experts?

RQ3: Do the modelers regard the DSL and the devel-

opment environment as user-friendly and learnable

for modeling CPs?

We answer RQ1 by identifying essential concepts

for modeling CP from literature research. Addi-

tionally, we selected the Smart Adaptive Case Man-

agement (SACM) (Michel, 2020), the engine of the

CONNECARE platform for integrated care that lever-

ages CP as a decision support tool for operating

treatments of chronic diseases (Vargiu et al., 2017).

SACM deﬁnes CPs using a custom XML structure,

thus we analyze the current XML syntax and SACM

capabilities to devise approaches for optimizing the

CP declaration with a DSL.

To model the CP concepts, we applied guidelines

and best practices of DSL design (Karsai et al., 2014,

pp. 3-5) to develop the Acadela syntax. First, our

DSL shall include only the necessary domain con-

cepts to model the identiﬁed CP requirements. Our

study develops a textual DSL because we aim to cre-

ate a lightweight, extensible DSL that is platform-

independent, i.e., the DSL can generate a CP model in

different formats (e.g., XML, UML) to be compatible

with various e-Health systems. To realize this goal,

we deﬁne the Acadela grammar using the textX meta-

language (Dejanovi

c et al., 2017). The textX gram-

mar parses the input CP model and generates the cor-

responding abstract syntax tree (AST). An Acadela

Interpreter processes the AST to construct a CP meta-

model object, and then translates the CP meta-model

into a SACM-compatible format, as shown in Fig-

ure 1. SACM creates CP models that execute patient

treatments at runtime from their meta-models.

Regarding usability, our DSL aims to provide low-

technical syntax, concise constructs, and ﬂexible syn-

tactic rules extracted from popular programming lan-

Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways

Figure 1: The process to generate an SACM CP meta-model from a CP deﬁnition code written in Acadela.

guages (e.g., no semicolon to end a statement, indent-

and case-insensitive like in SQL). We select these fea-

tures with the expectation that modelers with basic

programming knowledge can quickly learn and apply

our DSL’s modeling concepts. Afterward, we follow

the guidelines of design science from Hevner et al.

(2004, p. 80) to begin the iterative design, develop-

ment, and reﬁnement cycle of our DSL.

Considering Acadela’s expressiveness, RQ2

states the potential of modeling CPs with different

complexity applied in various medical departments.

Following this direction, we contacted six medical

experts and collected ﬁve CPs from their ﬁelds of ex-

pertise. We modeled the CPs in Acadela, which then

generates CPs in a format compatible with SACM.

We demonstrated the execution of CP models to

medical professionals in SACM to assess the DSL’s

accuracy and applicability from their perspective.

Finally, RQ3 explores the DSL’s usability from

the perspective of modelers. Thus, we conducted de-

scriptive user studies with technical staff in the health-

care sector or medical researchers with programming

or modeling experience. The participants developed

a hypertension CP using our Acadela IDE and gave

us quantitative and qualitative feedback on the DSL’s

user-friendliness, learnability, and applicability.

3 RELATED WORK

Graphical DSLs. Several DSLs are developed to

model CPs supporting evidence-based decisions in

clinical treatments. The motivation is that standard

business process modeling notations, like BPMN, do

not fully support modeling clinical procedures that

require discretionary execution paths to handle un-

predictable events. In this direction, Burwitz et al.

(2013) developed CP-Mod to support Evidence-based

Medicine (EBM) with the modeling of 1) decision

support, i.e., the recommended actions based on deci-

sion criteria, along with references to medical guide-

line documents; 2) Treatment alternatives and prob-

ability to indicate possible care approaches; 3) time

event representing time slots, waiting period or tem-

poral dependencies between activities.

Similarly, Braun et al. (2014) and Neumann et al.

(2016, 2017) extended BPMN to develop BPMN4CP

and BPMN

SIX

. Both DSLs ﬁrst construct a domain

ontology through 1) requirement analysis and 2) do-

main analysis to examine whether a modeling lan-

guage is appropriate to model the concepts, proper-

ties, and constraints of the domain. Next, they per-

form an equivalence check to assess which model-

ing concepts are constructible with default BPMN

elements and to which extent. After that, the au-

thors classify whether a domain concept is mode-

lable by combining multiple BPMN elements (Equiv-

alence by Composition) or using BPMN elements and

domain-speciﬁc attributes (Equivalence by Speciﬁca-

tion). The authors then apply a BPMN extension

procedure of Stroppi et al. (2011) to develop a do-

main class model (CDME) to extend concepts not

modelable by standard BPMN elements. The CDME

model is an abstract syntax that is derivable to con-

struct the concrete syntax (graphical notations) in a

BPMN extension model format (BPMN+X). As a re-

sult, BPMN4CP and BPMN

SIX

create an extended

BPMN model for CPs that includes quality indicators

and management of human, equipment, consumables,

and document resources concepts.

Textual DSL. Msosa (2018, 2019) developed

FCIG, a textual DSL to model Clinical Practice

Guidelines (CPGs) in Computer-Interpretable Guide-

lines (CIGs) forms. First, Msosa analyzed the domain

concepts of a CIG Guideline, which comprises a set

of Recommendations containing Actions and Condi-

tions to specify which medical interventions apply to

particular circumstances. Next, Msosa leveraged the

Xtext (Eclipse Foundation, nda) DSL build tool to de-

ﬁne FCIG grammar. When processing a Guideline,

FCIG navigates through the AST to 1) retrieve all

Recommendations in the Guideline, 2) for each Rec-

ommendation, return the list of Actions that satisfy the

ICSOFT 2023 - 18th International Conference on Software Technologies

Table 1: Features Comparison of DSLs for CP modeling.

F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13 SF1 SF2

CP-Mod

DSML4CPs

BPMN4CP

BPMN

SIX

FCIG

Acadela

No Support Partial Support Full Support

Condition, and 3) generate a function that evaluates

all Guideline’s Recommendations and returns appli-

cable Actions. The advantages of FCIG are 1) concise

syntax and 2) offering an IDE with syntax highlight-

ing and auto-completion to support modeling.

While existing DSLs model key aspects of CP op-

eration, they do not mention the support of external

systems communication, dynamic graphical render-

ing, and importing CP elements. Our study discusses

the necessity of the features, coded F11, F12, and F13

in Section 4, and our modeling solutions in Section 5.

4 CP MODELING FEATURES

From the literature study of existing DSLs, e-Health

systems, and CP deﬁnition and execution, we col-

lected the requirements for modeling executable CPs

in e-Health applications. Furthermore, we also collect

features for textual DSLs to support technical staff in

constructing CP models. Table 1 summarizes the im-

plementation of identiﬁed CP modeling (F#) and sup-

porting features (SF#) in DSLs from relevant litera-

ture and Acadela.

F1: The DSL shall model medical processes us-

ing concepts from the healthcare domains familiar to

healthcare professionals. Rationale: First, construct-

ing activities and phases of medical procedures is es-

sential to declare structured or alternative tasks in the

treatment. Second, concepts used in the DSL should

map to the terminology which prospective users are

familiar with (Frank, 2013, p. 136).

F2: The DSL shall model alternative treatment pro-

cesses, their triggering conditions, and potential val-

ues that trigger other alternatives. Rationale: Care

professionals frequently handle unpredictable scenar-

ios during the planned treatment process. Thus, a

DSL shall provide decision support to care profes-

sionals, patients, and their relatives (De Bleser et al.,

2006, p. 562).

F3: The DSL shall model evidence classes with data

assigned to them. Additionally, the DSL shall support

referencing the evidence. Rationale: CP processes

and decisions following EBM principles (Wang et al.

(2011, 2021)) shall be supplemented with informa-

tion on the evidence to decide the appropriate inter-

ventions.

F4: The DSL shall model iterative treatment pro-

cesses. Parallel repetition should also be supported.

Rationale: CPs shall support control ﬂow by exe-

cuting both optimal, sequential procedures and repet-

itive, unpredictable processes. Parallel execution is

necessary to construct simultaneous tests (e.g. paral-

lel PCR tests (Berdal et al., 2008), (Perchetti et al.,

2020, p. 2)).

F5: The DSL shall model each medical process’s

treatment goal(s) based on medical practices and pref-

erences of individual patients. Rationale: Express-

ing care goals based on the applied remedies and the

patient’s preferences fosters treatment transparency

(Heß et al., 2015, p. 6). Furthermore, this feature en-

courages medical professional teams to establish uni-

ﬁed remedies to achieve the care goal(s).

F6: The DSL shall support assigning responsibili-

ties to medical processes. The responsibilities distin-

guish between staff that executes medical tasks and

staff responsible for the treatment outcome. Ratio-

nale: Specifying the responsibilities of each health-

care professional optimizes the sequencing of medi-

cal procedures, indicates contact persons, and fosters

communication among staff (Heß et al., 2015, p. 6).

F7: The DSL should model checklists associated with

medical processes. Rationale: Checklists are widely

accepted and helpful instruments to present details

about the execution of medical processes (Healey

et al., 2011, p. 3) (Wolff et al., 2004, pp. 430-431).

F8: The DSL should model various aspects and infor-

mation needs of stakeholders in medical institutions.

Rationale: CPs aim to optimize medical care from

various perspectives. Therefore, a DSL should pro-

vide concepts to model treatment-related information,

such as (multidisciplinary) medical or organizational

aspects (Heß et al., 2015, p. 6). Besides, the DSL

should include medical guidelines to assist care pro-

Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways

fessionals to decide on appropriate interventions.

F9: The DSL should model the allocation of re-

sources to the medical process. Rationale: CPs also

aims to manage and optimize resource consumption

(Heß et al., 2015, p. 6). Resources comprise equip-

ment or documents used during the medical process

(Neumann et al., 2016, p. 4), medicine, and facilities

(Braun et al., 2016, p. 3252).

F10: The DSL should model time constraints and ex-

plicit time events. Rationale: the deﬁnition of CP in-

corporates the timing necessary to achieve care goals

(De Bleser et al., 2006), (Campbell et al., 1998), (Van-

haecht et al., 2006). Therefore, a DSL should provide

features to declare due dates or timed activities such

as periodic or time-lapsed tasks.

F11: The DSL should model external requests to or-

chestrate and integrate with existing or external sys-

tems (Michel, 2020, p. 37). Rationale: e-Health ser-

vices may need to interact with legacy systems that

store user data. Furthermore, delivering care services

may require sharing data with partner systems at dif-

ferent process lifecycles (e.g., activated, terminated).

For example, sending blood pressure data to a moni-

toring application when completing the measurement.

F12: The DSL should support customizing the graph-

ical representation of activity. Rationale: Customiz-

able UI display enhances extendability and supports

the modeling of specialized clinical use cases be-

cause modelers can deﬁne dynamic visual representa-

tions of data without relying on predeﬁned templates

(Michel and Matthes, 2018). Besides, a comprehen-

sive overview of medical status inﬂuences the anal-

ysis of the patient’s condition and decision-making

process of medical experts (Suganthi and Poongodi,

2021), which affects the treatment quality. Thus, a

DSL shall assist modelers in designing comprehen-

sive graphical representations of data.

F13: The DSL should support importing CP elements

into the current CP deﬁnition. Rationale: Repeatedly

declaring common procedures (e.g., patient admis-

sion) in each CP results in duplication and cumber-

some modiﬁcation for all affected CPs when the tasks

change. Therefore, deﬁning these shared elements

once and importing them into CPs foster reusability

and ﬂexibility of the modeling process.

SF1: The DSL should offer an IDE. For textual DSLs,

the IDE should have syntax highlighting and auto-

completion capabilities. Rationale: The combination

of auto-completion, syntax highlighting, and syntax

checking can enhance productivity when deﬁning CP

elements, increase usability (Cook et al., 2007, p. 16-

17) (Merkle, 2010), and prevent typos.

SF2: The DSL should visualize the modeled CP

to demonstrate the reconstructed elements, such as

phases, activities and their inputs or outputs, responsi-

bilities, and control ﬂows, using basic graphical nota-

tions. Rationale: A CP visualization provides visual

clues that can enhance the understandability and read-

ability of the models (Frank, 2010, p. 1). Another

beneﬁt is reducing the learning effort as the clinical

experts do not have to understand modeling concepts

in-depth (e.g., (Hermans et al., 2009, p. 433)). Addi-

tionally, one noticeable feedback from our interviews

with technical staff is that Acadela needs a graphical

visualization to provide an overarching overview of

the CP model. These visual clues assist the debug-

ging and analysis of the model.

Observation: Existing DSLs satisfy the F1 feature

by providing essential constructs to model generic CP

process elements, e.g., Activities (Braun et al., 2016,

p. 3252), ProcessDeﬁnition (Michel, 2020, p. 45).

In addition, the DSLs include concepts representing

decision criteria with conditional expression for acti-

vating alternative treatment paths (e.g., DecisionCri-

terion (Braun et al., 2014; Burwitz et al., 2013), Cri-

terion (Heß et al., 2015, p. 10), and SentryDeﬁnition

(Michel, 2020, pp. 45, 47)), thus supporting F2.

Most DSLs support features F3, F4-F6, and F8.

We considered FCIG to fulﬁll F8 partly because the

DSL can use the Action concept to open a medical

document at a particular URL. However, FCIG does

not offer constructs to model organizational aspects

(e.g., medical departments or teams). Most DSLs sup-

port modeling evidence classes and repetitive activ-

ities (F3 and F4). Furthermore, existing DSLs can

model resource management aspects in CP operations

(F6 and F8). Most DSLs can express treatment goals

(F5), thus assisting care professionals in communicat-

ing and planning their therapies. Not all DSLs support

features F7 and F10 because of the original scope of

their CP application context.

Regarding the supporting features SF1 and SF2,

graphical DSLs offer an expressive IDE with ex-

tended concepts to help modelers construct CPs by

interacting with graphical elements. FCIG, as a tex-

tual DSL, also develops an IDE with syntax highlight-

ing, yet it does not auto-complete keywords or visu-

alizing CIG models. Acadela auto-completion covers

keywords but not element IDs in a CP meta-model.

The F11, F12, and F13 features are the novel

contributions of Acadela, which aim to support inter-

system communications and provide ﬂexible model-

ing capabilities to modelers. Existing DSLs did not

discuss the three requirements, yet they are essential

in modeling CPs in integrated, adaptive e-Health plat-

forms as explained in the Rationale.

The following section presents our methods to

ICSOFT 2023 - 18th International Conference on Software Technologies

model CP that support the 15 identiﬁed features.

5 CP METAMODEL DEFINITION

In SACM, medical experts and modelers need to col-

laborate to deﬁne Stages and Tasks in the care pro-

cess, control ﬂow, responsibilities, and communica-

tion to external systems in a CP. The ﬁrst three ele-

ments are foundational building blocks for develop-

ing standardized and multidisciplinary CPs. This sec-

tion presents the Acadela grammar to enforce syntac-

tic rules of CP declaration, followed by the CP mod-

eling concepts and their concrete syntax in Acadela to

address the CP modeling requirements.

5.1 Grammar

Acadela arranges CP elements in a subsumption order

to provide a hierarchical structure of CP models. For

example, a Case object contains multiple Stages and

each Stage consists of Tasks. Each CP element has

user-deﬁned attributes following the key=value pat-

tern. In addition, Acadela expresses attributes accept-

ing only SACM-deﬁned values as directives, which

conforms to the #<attributeValue> pattern. The be-

low code expresses the semantics of Acadela gram-

mar in Extended Backus Naur Form (EBNF). To con-

dense the grammar, attributes, and directives of a

CP element are declared as <CpElement>Attrs or

<CpElement>Drts, respectively.

In EBNF, the # sign denotes the order of rules

within a group is interchangeable. The * sign means

a CP element has zero or more instances, while the +

sign allows one or more instances. The ? mark repre-

sents an optional element.

Sta rt = ( Im port *) ? ( Wo r k sp ac e | O b j )

Wo rk sp ac e = " Wor ks pa ce " ID

" d ef in e " C ase

Obj = Ca se | C a s eS et t i n g | Sta ge

| T ask | Form | A t tr i bu t e V al u e

| Ou t p u tF ie l d | I n p ut Fi el d | Ho oks

Hook = H tt pH oo k | Ca se Ho ok

| Du a lT as k Ho ok

Case = " Cas e " ID ( Cas eA t t r s C a s eS et t i n g

Re s p o n s i bi l it i es ( Ca se Ho ok +) ?

Su mm a r y Pa n e l S tag e +) #

Re s p o n s i bi l it i es = ( G rou p * U ser *) #

Ca se S e t ti ng = ( Ca se Ow ne r C as e P a ti en t

Ca s e S e t t in g At t rs ) #

Sta g e = " S ta g e " ID W or kf l ow Dr t s

St ag eA t t r s

( Ht tp Ho ok * Pr e co nd i ti on * Task +) #

Task = H um an Ta sk | Du al Ta sk

| Au t om a t e dT a sk

Hu ma nT as k = " Hum an Ta sk " ID W o rk f l o wD r t s

( Ta s k A tt rs Pr ec o nd it i o n *) #

( Ht tp Ho ok * F orm )#

Form = " For m " ID ( For mD rt s )? Fie ld +

Fie l d = I n p ut Fi el d * O u tp ut F i e ld *

In pu tF i e l d = " Inp ut Fi el d " ID

In p u t Fi e ld D r t s

In p u t Fi e ld A tt r s

Ou tp u t F ie ld = " Ou t pu tF i e l d " ID

Ou t p u tF i el d Dr t s

Ou t p u t F i el d At t rs

Ou t p u tF i el d Dr t s = ( Man da to ry R e a d o n l y )#

Ma nd at or y = "#" " ma nd at or y "

| " no t Ma n d a to r y "

Ou t p u t F i el d At t rs = ( De s cr ip t io n

Ex pr es s i o n ...) #

De sc r i p ti on = " l abe l " "=" ST RI NG

Regarding Tasks, since the syntax of HumanTask,

DualTask, and AutomatedTask are similar, the gram-

mar only expresses HumanTask. Similarly, Output-

FieldDrts and OutputFieldAttrs characterize the syn-

tax of directives and attributes of other CP elements.

5.2 Modeling CP Elements

Care Process (F1, F3, F4, F7, F12). In SACM, a

Workspace represents the medical institution provid-

ing healthcare services, and a Case is the CP for treat-

ing a particular disease. Acadela expresses treatment

phases and activities in a CP using Stages and Tasks

(F1). Since clinical processes are multi-stage proce-

dures, they can comprise distinct and possibly repet-

itive phases. SACM supports a parallelly repeated

Stage (Michel, 2020, pp. 80, 105), which Acadela

expresses as a #parallelRepeat directive (F4). Each

Stage comprises Tasks of type: human, automated,

or dual tasks, which are, respectively, conducted by

a person, third-party system, or both consecutively.

Listing 1 shows the Acadela syntax to declare a re-

peated Stage and a HumanTask to measure the Body

Mass Index (BMI) value. Figure 2 displays the result-

ing UI of Listing 1.

Sta g e E va lu at i o n #r ep e at Se r ia l

lab e l = ’ Eva lu at io n ’

Hu ma nT as k M ea su r e B mi

lab e l = ’ Me as ur e B MI ’

du eD at e R e f = ’ Se tt in g . Du eD at e ’

Form B mi Fo r m

In pu tF i e l d He igh t # nu mbe r (0 -3)

lab e l = ’ He ig h t ( m ): ’

In pu tF i e l d We igh t # nu mbe r (0 -300)

lab e l = ’ We ig h t ( kg ) : ’

Ou tp u t F ie ld Bm iS co re #n u mb er

lab e l = ’ BMI Sc ore : ’

Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways

ex pr es s i o n =

’ We ig ht / ( He ig ht * H e ig ht ) ’

uiR e f = ’ co lo r s (0 < ye ll o w < 18.5

< gr een < 25 < red < 10 0) ’

Listing 1: Sample deﬁnition of a BMI Measurement Task

in an Evaluation Stage.

Figure 2: The UI output from the Stage Declaration in List-

ing 1. The Figure shows the activated Evaluation Stage and

enabled MeasureBMI Task (left), and the UI of the Task

(right).

Each Task has a Form containing InputFields to

accept medical data to collect necessary medical con-

ditions for an evidence class, diagnosis, or documen-

tation (F1, F3). The medical information supports

care experts in determining the appropriate interven-

tions in EBM. Besides, Acadela represents a checklist

(F7) by deﬁning multiple-choice InputFields that ac-

cept multiple answers. Additionally, the OutputFields

can visualize the input data in the SACM UI (F12).

Control Flow (F2): Medical experts execute alter-

native actions depending on the patient condition.

Therefore, CPs shall support conditional activation of

Stages or Tasks depending on speciﬁc medical data.

Listing 2 demonstrates the syntax of deﬁning transi-

tion conditions using the Precondition construct in-

side a Stage. The previousStep attribute states the pre-

requisite Stage or Task, while the condition attribute

is a boolean expression for activating the Stage.

Sta g e E va lu at i o n #r ep e at Se r ia l

... // S t age At tr ib ut e s

Pr ec o n d it i o n

pr ev i o u sS t e p = ’ I d en t if i c a ti o n ’

co nd it io n = ’ Id e n ti f ic a t i on .

Fi l l C on s en t Fo r m . Con s en t = 1 ’

Listing 2: Deﬁning a transition condition of a Stage. Here

the Evaluation Stage is activated if the Identiﬁcation Stage

is completed and the patient consents with the treatment.

CP Summary Panel (F5): Listing 3 demonstrates

a SummaryPanel code displaying the treatment goal

from an InputField. This element displays the value

from an InputField or OutputField in the Case Sum-

mary page. InputFields can record the care goal while

Su mm a r y Pa n e l

Se c ti on Ta rg et Bm i # left

lab e l = " Ta rg e t BMI : "

In fo P a t h I d en t if i c a ti o n .

Se t T r e a t me n tG o al . Tar ge tB mi

Listing 3: SummaryPanel Deﬁnition to show the care goal.

Acadela traces the InputFields from the InfoPath attribute

by parsing the <StageId>.<TaskId>.<FieldId> pattern.

the OutputFields values indicate the progress or con-

straints to achieve the goal.

Responsibilities (F6). SACM supports referenc-

ing individual experts or professional groups in the

database and grants them access rights for operating

Stages or Tasks (F6), as shown in Listing 4. Model-

ers can optionally assign the unique ID of a Group or

User to the staticId attribute.

Re s p o n s i bi l it i es

Gro u p S tP a u l Nu r s e s name = ’ N u rs es ’

st at i c I d = ’ ef d c 39 5 2 e s7 5 ’

Se t ti ng

At tr ib ut e Nu rs es

#Link .#Us ers ( St P au l N u rs e s )

lab e l = ’ Nu rs e s ’

...

Sta g e E va lu at i o n

own e r = ’ Se tt in g . Nu rse s ’

Listing 4: SummaryPanel Deﬁnition to show the care goal.

Document (F8). In case medical documents (e.g.,

guidelines, forms) necessary to provide care services

(Heß et al., 2015; Braun et al., 2014) are accessible via

URLs, modelers can declare a link to each document,

as shown in Listing 5.

In pu tF i e l d M ed i ca l Gu i de l i n e

#d oc u m e nt L i n k (’ ht tps :// s tp a ul . de /

fi le p a t h ’)

lab e l =" L ink to Me di ca l Gu i d e l i ne :"

Listing 5: Deﬁning a link to a document in Acadela.

Due Date Deﬁnition (F10). SACM supports deﬁn-

ing a default due date from the current time in the

Case Setting. The due date is assignable to a Human-

Task or a DualTask, as shown in Line 6 of Listing 1.

External System Communication (F11). SACM

updates medical status in external systems by sending

HTTP requests (HttpHook) with Case, Stage, or Task

data when an event (e.g., activated, completed) oc-

curs in these elements (Michel, 2020, p. 113). Listing

6 deﬁnes a HttpHook inside the MeasureBMI Task.

Upon the Task completion, the HttpHook sends the

ICSOFT 2023 - 18th International Conference on Software Technologies

Figure 3: Illustration of selected massage positions in an InputField (left) affect the output image visualization (right).

Task data from its InputFields and OutputFields to

the external system URL using the POST method. If

the request fails, SACM shows the value of the fail-

ureMessage attribute to the user.

Hu ma nT as k M ea su r e B mi

lab e l = ’ Me as ur e B MI ’

... // O t her T a sk ’s el em en ts

Tr i gg er

On c om pl et e in v ok e ’ ht t ps ://

pa rt n er s y s te m . de / up da te ’

met ho d P OST wit h fa i lu re M es s ag e

" C an no t upd at e i nfo !"

Listing 6: Deﬁning an external request triggered upon a

Task completion

Customizable Graphical Representation (F12).

The SACM UI of a Task is a grid layout of three

columns. Acadela speciﬁes the column position of an

InputField or OutputField using a directive, e.g., #left,

#center, #right (F12). Besides, SACM supports ap-

plying color to a range of numeric values. Line 17 in

Listing 1 shows how Acadela deﬁnes the green color

of the BMI OutputField for a value from 18.5 to 24.9.

To render a customizable UI Template, an Out-

putField contains 1) a SVG image comprising graph-

ical elements to visualize medical data, 2) the CSS

style declaring the UI effect applied to the elements

(e.g., hide, show, fore- or background color), and 3)

a SACM command to dynamically modify the CSS

style based on InputField(s) value. Figure 3 shows an

example of displaying massage positions on a human

head based on the selected areas.

Import CP Element (F13). Acadela leverages the

textX scope provider to import CP elements deﬁned

in a ﬁle into the current CP. textX supports FullyQual-

iﬁedName (FQN) scope that expresses the location of

an imported object based on the ﬁle path and object

ID (Dejanovi

c, ndc). For instance, Listing 7 shows

how to import a Discharge Stage deﬁned in the dis-

charge.aca ﬁle under the stage folder into a CP model.

// Fil e p a th : ./ st a ge / d is ch ar g e . ac a

def in e Sta g e D i s ch ar ge

... // S t age At tr ib ut e s

// - - - - - - -- - - - - -

imp or t sta ge s . di sc ha r g e

// im po rt s t ag es . d is ch ar g e as di

Wo rk sp ac e S tP a ul Cl i ni c

def in e C ase O T1 _ O b es it y

...

Sta g e E va lu at i o n ...

use S tag e Di s c h a r g e

// use St age di . Dis ch ar g e

Listing 7: Importing the Discharge Stage (lines 1-4) into

a CP. Lines 6 and 12 show how to import an element as

alias.

Web-Based IDE (SF1). To enhance the usability

and productivity of modelers, we developed a Web

IDE powered by Monaco

with 1) an auto-complete

feature to create CP elements from templates and 2)

syntax highlighting to help modelers identify CPs el-

ements.

CP Visualization (SF2). Acadela leverages GoJS

to develop a CP visualizer. GoJS provides functions

to graphically present CP elements. From the JSON

CP model generated by the backend, Acadela extracts

each CP element type and attribute to construct nodes

in the graph. Figure 4 shows the GoJS visualization

of a Hypertension CP used in our usability evaluation.

The Acadela Wiki

documents the DSL syntax for

deﬁning CP elements.

6 EVALUATION

To explore the applicability of Acadela, we designed

two descriptive surveys to assess 1) expressiveness,

i.e., the ability to model CPs in different medical

ﬁelds, and 2) usability from the perspective of model-

ers. We recruited voluntary participants from staff

working in medical facilities or research institutions.

For each evaluation, we collected quantitative data by

a survey and applying a Likert scale from 1 (Strongly

Disagree) to 5 (Strongly Agree) for each statement.

We also interviewed participants to obtain feedback

regarding the DSL’s applicability and limitations.

https://microsoft.github.io/monaco-editor/

https://gojs.net/latest/

https://cobalt-plot-a08.notion.site/Acadela-Wiki-

ee58b3f2eb6a4627b55b470fe1717d9f

Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways

Figure 4: The visualization of the simpliﬁed Hypertension CP.

6.1 Expressiveness

Population Sampling: We contacted 16 medical

professionals in Germany and received six responses.

The response rate is moderate at 37.5 percent. The

age range of participants is from 20 to 60 years old.

Two medical professionals have more than 25 years

of experience, while others have up to ﬁve years.

Study Design: The six medical professionals

shared ﬁve CPs for treating Chronic Obstruction Pul-

monary Disease (COPD), Schizophrenia, Chronic

Headache, and two Diagnoses of Smoke Inhalation

Injury and Cervical Cancer. The CPs have various

complexity ranging from linear to dynamic, adaptive

treatment procedures. The complex CPs demonstrate

the language ability in modeling transition conditions

for unpredictable, personalized treatment processes.

The ﬁrst three CPs are linear workﬂows, while the

other two support the decision-making with transi-

tion conditions applied to input medical data. The

last two CPs are complex due to their adaptive and

non-deterministic nature.

Experiment Environment: We host Acadela and

SACM in our computing devices and show them of-

ﬂine or share the screen control via Zoom to avoid

being blocked by strict ﬁrewall policies in medical in-

stitutions. Due to the Corona pandemic, depending

on the participant’s preference, we presented the CP

model, collected the survey, and interviewed them on-

site or via Zoom.

Result: For each participant, we model their CP and

presented it in the SACM web application at their

workplace or via Zoom. After seeing the CP execu-

tion, we share the survey link with the participants and

ask for their opinions. Figure 5 shows the assessment

of medical experts on the accuracy of the modeled CP.

Figure 5: Medical professionals evaluation regarding the

accuracy of the modeled CPs.

The result suggests that the system constructs CPs

accurately with all participants agreed or strongly

agreed with each statement. In other words, Acadela

is potentially capable of reconstructing various medi-

cal procedures precisely, such that the SACM system

can correctly show and execute the treatment process

accordingly. Furthermore, in the subsequent inter-

view, the participants considered the modeled CPs as

correct. However, our Cervical Cancer Diagnosis CP

model does not correctly show the period before the

next test. Regarding the applicability, the participants

regarded the CP models as ”appropriate as a guide-

line”. Nevertheless, they should ”address treatment

variations” to be practically applicable. The evalua-

tion outcome implies a threat to internal validity, as

our study focuses on standard treatment processes but

did not substantially model potential alternative paths

in a CP. However, medical experts consider modeling

variations as a signiﬁcant indicator of expressiveness.

When considering external validity, the limited

number of CPs and involved medical professionals in

our study suggests that Acadela may need additional

ICSOFT 2023 - 18th International Conference on Software Technologies

modeling concepts to express the medical terminol-

ogy and procedures in other treatments.

6.2 Usability

Population Sampling: We invited 21 technical

staff at medical or research institutions and recruited

eight participants from Germany, Spain, the Nether-

lands, and Italy. The response rate is moderate at 33.3

percent. Seven participants were 18 to 45 years old,

and one technical expert is over 46. Three partici-

pants have less than four years of experience, others

have more than eight years.

Study Design: All participants joined the experi-

ments online via Zoom. Each participant ﬁrst at-

tended a training session, in which we introduced

how Acadela models CP elements and practiced

deﬁning a CP together by completing an exercise.

Next, in a modeling session, the participants con-

structed elements of an incomplete hypertension CP

and ﬁxed bugs of another CP model by examining

their error messages (EMs). We controlled the eval-

uation tasks and bugs such that all participants mod-

eled the same CP elements and debugged the same

syntax and semantic errors. Performing the same ac-

tivities builds a common baseline to gauge usability

and learnability.

Result: After the debugging session, we asked the

experts to assess the usability of Acadela and its er-

ror validation. We presented a questionnaire with

statements regarding the a) User-friendliness of the

DSL, b) Helpfulness of EMs, and c) Overall usability

assessment using the System Usability Scale (SUS).

Figure 6 presents the evaluation outcome.

In the subsequent interview, overall, the partici-

pants comment on the syntax as ”simple”, ”elegant”,

and ”easy to learn” for users with competent program-

ming experience. In addition, EMs are considered

”useful” as they ”show the line number and the prob-

lem” to help users ”pinpoint the error”. Six out of

eight experts considered Acadela to have the poten-

tial to model CPs in e-Health applications.

Regarding limitations, the auto-complete feature

should ”show the names of elements in the code”,

which is convenient for the users and prevents typ-

ing a non-existing CP component. Furthermore, three

participants share an opinion that Acadela is ”fairly

easy to use”, but ”depend on the background and ed-

ucation” of modelers. Therefore, modelers need fun-

damental programming skills to model and debug the

Acadela code. One participant states that Acadela

ID Syntax Usability Statement

S1 The syntax for creating CP elements, i.e. Stage,

Task, Form, Field, is straightforward

S2 Editing CP elements is easy with the language

S3 Importing external modules is straightforward

S4 The language syntax was easy to learn and use

Figure 6: Usability quantitative evaluation result.

should model ”infographic”, ”videos”, and ”gamiﬁ-

cation factors” to further support medical staff.

7 CONCLUSION

Our study develops Acadela, a textual DSL for mod-

eling CPs with visualization capability. Acadela sup-

ports the deﬁnition of workﬂow phases and activ-

ities, control ﬂow, responsibilities, medical docu-

ment resources, timing constraints, communication to

external services, dynamic graphical representation

of data and the import of these elements to foster

reusability (RQ1). The language is overall consid-

ered user-friendly and learnable by professionals with

technical knowledge.

We leveraged the textX meta-language, guide-

lines of DSL design, and error messages to develop

the grammar, concrete syntax, and error validator of

Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways

Acadela. To evaluate our DSL expressiveness (RQ2),

we modeled ﬁve CPs and presented the executable

workﬂow in SACM, an e-Health platform for inte-

grated care, to six care professionals in ﬁve medical

ﬁelds. The result suggests that Acadela can accu-

rately model CP workﬂows in different medical de-

partments and complexity. However, both medical

professionals and technical staff wished to see more

modeling of visualization and variations in their CP.

To evaluate Acadela’s usability (RQ3), we invited

eight technical experts working in the healthcare in-

dustry or research to model a simpliﬁed CP for hyper-

tension treatment and ﬁx bugs of another CP based on

their error messages (EMs). The result implies that

Acadela is potentially usable and learnable to tech-

nical staff. Most participants deem the syntax ”sim-

ple” and ”easy to learn”, while EMs help them lo-

cate, understand and ﬁx errors. Therefore, they con-

sider the DSL applicable to model CPs in e-health ap-

plications. However, we reﬂected on the result that

Acadela is user-friendly for modelers with solid basic

programming knowledge or above. In addition, the

participants wished to see more capabilities, such as

the rendering of statistics. and auto-completion of CP

elements’ ID. Another concern is more participants

are needed to assess the DSL usability substantially.

Future Work. To further validate the applicability

of Acadela, one can expand the evaluation for mod-

eling CPs in more medical ﬁelds, such as Chinese

medicine or Orthopedics. Operating CPs in practi-

cal settings enable us to address the daily-life needs

and concerns of medical professionals while address-

ing the unique challenges of applying CPs in the med-

ical ﬁeld. Furthermore, different medical treatments

require diverse forms of medical data visualization

and additional medical concepts, hence we can inves-

tigate the applicability and limitation of CP modeling

features provided by Acadela. This future practical

study is the ﬁrst cornerstone for identifying potential

extensions or alternative solutions to assist medical

professionals in delivering quality care to patients.

ACKNOWLEDGEMENTS

We sincerely thank the medical professionals, re-

searchers, and technical staff for their time, support,

and patience in providing clinical pathways, conduct-

ing experiments, and valuable feedback to our study.

REFERENCES

Baar, T. (2015). Veriﬁcation support for a state-transition-

dsl deﬁned with Xtext. In International Andrei Er-

shov Memorial Conference on Perspectives of System

Informatics, pages 50–60. Springer.

Berdal, K. G., Bøydler, C., Tengs, T., and Holst-Jensen, A.

(2008). A statistical approach for evaluation of PCR

results to improve the practical limit of quantiﬁcation

(LOQ) of GMO analyses (SIMQUANT). European

Food Research and Technology, 227(4):1149–1157.

Braun, R., Schlieter, H., Burwitz, M., and Esswein, W.

(2014). Bpmn4cp: Design and implementation of

a bpmn extension for clinical pathways. In 2014

IEEE international conference on bioinformatics and

biomedicine (BIBM), pages 9–16. IEEE.

Braun, R., Schlieter, H., Burwitz, M., and Esswein, W.

(2016). BPMN4CP Revised–Extending BPMN for

Multi-perspective Modeling of Clinical Pathways. In

2016 49th Hawaii International Conference on Sys-

tem Sciences (HICSS), pages 3249–3258. IEEE.

Burwitz, M., Schlieter, H., and Esswein, W. (2013). Mod-

eling clinical pathways-design and application of a

domain-speciﬁc modeling language.

Campbell, H., Hotchkiss, R., Bradshaw, N., and Porte-

ous, M. (1998). Integrated care pathways. Bmj,

316(7125):133–137.

Cook, S., Jones, G., Kent, S., and Wills, A. C. (2007).

Domain-speciﬁc development with visual studio dsl

tools. Pearson Education.

De Bleser, L., Depreitere, R., Waele, K. D., Vanhaecht, K.,

Vlayen, J., and Sermeus, W. (2006). Deﬁning path-

ways. Journal of nursing management, 14(7):553–

563.

Dejanovi

c, I., Vaderna, R., Milosavljevi

c, G., and Vukovi

Z. (2017). TextX: a Python tool for domain-speciﬁc

languages implementation. Knowledge-based sys-

tems, 115:1–4.

Dejanovi

c, I. (n.d.c). Reference resolving expression lan-

guage (rrel). http://textx.github.io/textX/3.0/rrel/.

Eclipse Foundation (n.d.a). Language engineering for ev-

eryone! https://www.eclipse.org/Xtext/.

Every, N. R., Hochman, J., Becker, R., Kopecky, S., and

Cannon, C. P. (2000). Critical pathways: a review.

Circulation, 101(4):461–465.

Frank, U. (2010). Outline of a method for designing

domain-speciﬁc modelling languages. Technical re-

port, ICB-research report.

Frank, U. (2013). Domain-speciﬁc modeling languages: re-

quirements analysis and design guidelines. In Domain

engineering, pages 133–157. Springer.

Hai, J.-J., Wong, C.-K., Un, K.-C., Wong, K.-L., Zhang, Z.-

Y., Chan, P.-H., Lam, Y.-M., Chan, W.-S., Lam, C.-

C., Tam, C.-C., et al. (2019). Guideline-based critical

care pathway improves long-term clinical outcomes in

patients with acute coronary syndrome. Scientiﬁc Re-

ports, 9(1):1–9.

Healey, A. N., Nagpal, K., Moorthy, K., and Vincent, C. A.

(2011). Engineering the system of communication

ICSOFT 2023 - 18th International Conference on Software Technologies

for safer surgery. Cognition, Technology & Work,

13(1):1–10.

Hermans, F., Pinzger, M., and Deursen, A. v. (2009).

Domain-speciﬁc languages in practice: A user study

on the success factors. In International Conference on

Model Driven Engineering Languages and Systems,

pages 423–437. Springer.

Heß, M., Kaczmarek, M., Frank, U., Podleska, L., and

ager, G. (2015). A domain-speciﬁc modelling lan-

guage for clinical pathways in the realm of multi-

perspective hospital modelling.

Hevner, A. R., March, S. T., Park, J., and Ram, S. (2004).

Design science in information systems research. MIS

quarterly, pages 75–105.

Iroju, O., Soriyan, A., Gambo, I., Olaleke, J., et al. (2013).

Interoperability in healthcare: beneﬁts, challenges and

resolutions. International Journal of Innovation and

Applied Studies, 3(1):262–270.

Jouault, F., B

ezivin, J., and Kurtev, I. (2006). TCS: a DSL

for the speciﬁcation of textual concrete syntaxes in

model engineering. In Proceedings of the 5th inter-

national conference on Generative programming and

component engineering, pages 249–254.

Karsai, G., Krahn, H., Pinkernell, C., Rumpe, B.,

Schindler, M., and V

olkel, S. (2014). Design guide-

lines for domain speciﬁc languages. arXiv preprint

arXiv:1409.2378.

Kurz, M., Schmidt, W., Fleischmann, A., and Lederer, M.

(2015). Leveraging CMMN for ACM: examining the

applicability of a new OMG standard for adaptive case

management. In Proceedings of the 7th international

conference on subject-oriented business process man-

agement, pages 1–9.

Merkle, B. (2010). Textual modeling tools: Overview

and comparison of language workbenches. In

Proceedings of the ACM International Confer-

ence Companion on Object Oriented Program-

ming Systems Languages and Applications Com-

panion, OOPSLA ’10, page 139–148, New York,

NY, USA. Association for Computing Machinery.

https://doi.org/10.1145/1869542.1869564.

Michel, F. (2020). A Collaborative Purely Meta-Model-

Based Adaptive Case Management Approach for Inte-

grated Care. Dissertation, Technical University Mu-

nich, Munich.

Michel, F. and Matthes, F. (2018). A holistic model-based

adaptive case management approach for healthcare. In

2018 IEEE 22nd International Enterprise Distributed

Object Computing Workshop (EDOCW), pages 17–

26. IEEE.

Mitchell, J. (1999). From telehealth to e-health: the un-

stoppable rise of e-health. Department of Communi-

cations, Information Technology and the Arts.

Msosa, Y. J. (2018). Modelling evolving clinical practice

guidelines: a case of Malawi. PhD thesis, University

of Cape Town.

Msosa, Y. J. (2019). FCIG grammar evaluation: A usability

assessment of clinical guideline modelling constructs.

In 2019 IEEE Symposium on Computers and Commu-

nications (ISCC), pages 1141–1146. IEEE.

Neumann, J., Rockstroh, M., Franke, S., and Neumuth,

T. (2016). BPMNSIX–A BPMN 2.0 Surgical In-

tervention Extension. In 7th workshop on model-

ing and monitoring of computer assisted interventions

(M2CAI), 19th international conference on medical

image computing and computer assisted interventions

(MICCAI 2016), Athens, Greece.

Neumann, J., Wiemuth, M., Burgert, O., and Neumuth, T.

(2017). Application of activity semantics and BPMN

2.0 in the generation and modeling of generic surgical

process models. Journal of the International Foun-

dation for Computer Assisted Radiology and Surgery:

International Journal of Computer Assisted Radiology

and Surgery, 12:48–49.

Panella, M., Marchisio, S., Barbieri, A., and Di Stanislao,

F. (2008). A cluster randomized trial to assess the

impact of clinical pathways for patients with stroke:

rationale and design of the Clinical Pathways for Ef-

fective and Appropriate Care Study [NCT00673491].

BMC health services research, 8(1):1–8.

Panella, M., Marchisio, S., and Di Stanislao, F. (2003). Re-

ducing clinical variations with clinical pathways: do

pathways work? International Journal for Quality in

Health Care, 15(6):509–521.

Perchetti, G. A., Nalla, A. K., Huang, M.-L., Jerome,

K. R., and Greninger, A. L. (2020). Multiplexing

primer/probe sets for detection of SARS-CoV-2 by

qRT-PCR. Journal of Clinical Virology, 129:104499.

Preston, S., Markar, S., Baker, C., Soon, Y., Singh, S., and

Low, D. (2013). Impact of a multidisciplinary stan-

dardized clinical pathway on perioperative outcomes

in patients with oesophageal cancer. Journal of British

Surgery, 100(1):105–112.

Rieger, C., Westerkamp, M., and Kuchen, H. (2018). Chal-

lenges and Opportunities of Modularizing Textual

Domain-Speciﬁc Languages. MODELSWARD, pages

387–395.

Stroppi, L. J. R., Chiotti, O., and Villarreal, P. D. (2011).

Extending BPMN 2.0: method and tool support. In

International Workshop on Business Process Model-

ing Notation, pages 59–73. Springer.

Suganthi, S. and Poongodi, T. (2021). Interactive Visualiza-

tion for Understanding and Analyzing Medical Data.

In Exploratory Data Analytics for Healthcare, pages

101–123. CRC Press.

Thakrar, B. T., Grundschober, S. B., and Doessegger, L.

(2007). Detecting signals of drug–drug interactions

in a spontaneous reports database. British journal of

clinical pharmacology, 64(4):489–495.

Vanhaecht, K., WITTE, K. D., Depreitere, R., and Sermeus,

W. (2006). Clinical pathway audit tools: a systematic

review. Journal of nursing management, 14(7):529–

537.

Vargiu, E., Fern

andez, J., Miralles, F., Cano, I., Gimeno-

Santos, E., Hernandez, C., Torres, G., Colomina, J.,

de Batlle, J., Kaye, R., et al. (2017). Integrated care

for complex chronic patients. International Journal of

Integrated Care, 17(5).

Wang, S., Yu, H., Liu, J., and Liu, B. (2011). Explor-

ing the methodology and application of clinical path-

Acadela: A Domain-Speciﬁc Language for Modeling Clinical Pathways

way in evidence-based Chinese medicine. Frontiers of

medicine, 5(2):157–162.

Wang, X., Chen, J., Peng, F., and Lu, J. (2021). Construc-

tion of clinical pathway information management sys-

tem under the guidance of evidence-based medicine.

Journal of Healthcare Engineering, 2021.

White, M. (2009). Case management: Combining knowl-

edge with process. BPTrends, July.

Wienands, C. and Golm, M. (2009). Anatomy of a visual

domain-speciﬁc language project in an industrial con-

text. In International Conference on Model Driven

Engineering Languages and Systems, pages 453–467.

Springer.

Wolff, A. M., Taylor, S. A., and McCabe, J. F. (2004). Us-

ing checklists and reminders in clinical pathways to

improve hospital inpatient care. Medical Journal of

Australia, 181(8):428–431.

Yang, W. and Su, Q. (2014). Process mining for clinical

pathway: Literature review and future directions. In

2014 11th international conference on service systems

and service management (ICSSSM), pages 1–5. IEEE.

ICSOFT 2023 - 18th International Conference on Software Technologies