Human-Centered e-Health Development: An Accessible Visual Modeling

Tool

Jingyuan Shen

, Hourieh Khalajzadeh

2 a

and Anuradha Madugalla

1 b

HumaniSE Lab, Monash University, Clayton, Victoria, Australia

School of Information Technology, Deakin University, Melbourne, Australia

Keywords:

Domain Speciﬁc Modeling Language (DSML), e-Health, Software Modeling, Visual Notation.

Abstract:

e-Health enables easy access to medical services without some of the limitations of traditional medical ser-

vices such as restricted access to specialists and the need to travel long distances. With the recent advances in

IT, more and more people are adopting e-health solutions. However, most of the domain experts and end users

who may be involved in developing these e-health applications may not have an IT background. This makes

it hard for them to contribute to its development. This paper presents a human-centered e-Health modeling

language, to help end-users easily specify their requirements and communicate with domain experts and clin-

icians to design and develop personalized e-health applications. The tool aims to provide a common language

between developers and domain experts, assisting the interdisciplinary teams to focus on the project itself

rather than on communication. We built a visual web application using the SiriusWeb platform to implement

our approach and evaluated its accessibility and usability with 11 end users. Participants reported that the tool

was straightforward to use without any IT knowledge, and notations were distinguishable and expressive.

1 INTRODUCTION

E-Health is “The use of information and communi-

cations technology in support of health and health-

related ﬁelds” (WHO, 2019). It enables people to ac-

cess a wide range of medical services from anywhere.

Compared to traditional health services, e-Health al-

lows to deliver medical services for less cost, espe-

cially when medical resources are limited.

As more people start accessing health-related in-

formation through the internet, the majority are seek-

ing for more personalized, more accessible, and more

interactive medical services to improve their and their

family’s health. This leads to the challenge of meet-

ing the needs of different patient groups. In order

to effectively address this people with diverse back-

grounds need to join in e-Health software develop-

ment projects. Such projects involve bringing to-

gether stakeholders with a variety of skill sets, di-

verse expertise, and backgrounds to communicate and

interact as multidisciplinary teams. In such teams,

many will not be experts in software programming

(Khalajzadeh et al., 2020b). Therefore, some soft-

ware engineering projects implement Model-driven

https://orcid.org/0000-0001-9958-0102

https://orcid.org/0000-0002-3813-8254

Approach (MDA) in order to address issues brought

by the interaction between multi-disciplinary stake-

holders (Fischer et al., 2014). MDA abstracts key

software components into models, fuses them with

identiﬁers and notations to create a ﬁctitious system

architecture, and tests the system’s performance be-

fore the development and deployment. The main ben-

eﬁt of using MDA in software engineering is that even

though some of the project stakeholders have no IT

background and relevant skills, they can still engage

in the project and provide their insights and opinions

based on their domain expertise (Zhuang et al., 2022).

Software engineering has several modeling lan-

guages, such as UML, SysML (Hause et al., 2006),

WebML (Ceri et al., 2000), etc. to help software engi-

neers develop software more efﬁciently. However, by

exploring and comparing existing research, we found

several gaps in the existing modeling languages, such

as accessibility issues, and a lack of systematic and

human-centered evaluation.

The main objective of our paper is to present a

new Domain Speciﬁc Modeling Language (DSML)

for designing and modeling e-Health applications.

We present this along with a visual-based human-

centered modeling tool that uses our DSML to engage

all stakeholders in e-Health development projects.

Shen, J., Khalajzadeh, H. and Madugalla, A.

Human-Centered e-Health Development: An Accessible Visual Modeling Tool.

DOI: 10.5220/0012546100003687

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 19th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE 2024), pages 283-290

ISBN: 978-989-758-696-5; ISSN: 2184-4895

283

Our new modeling tool deﬁnes key components of

e-Health and provides visual notations based on the

Web Content Accessibility Guidelines (WCAG) (An-

drew et al., 2018). It assigns corresponding symbols

to all concepts and allows users to select their pre-

ferred visual design. WCAG provides a wide range

of recommendations for improving the accessibility

of visual elements (Andrew et al., 2018), and these

helped us to design our visual notations. With this

process we have contributed to making e-Health de-

velopment processes more accessible and efﬁcient to

multiple stakeholders. We evaluated our visual no-

tations by applying the rules of Physics of Notations

(PoN) (Moody, 2009) to the notations and by conduct-

ing a user study with a group of 11 end-users includ-

ing software engineers and business analysts.

2 MOTIVATION

MDA is used in software engineering to provide

guidelines and speciﬁcations for projects and their ﬁ-

nal deliverable, making it one of the most efﬁcient

ways of developing software (Brambilla et al., 2017).

In the beginning of a typical software engineering

project, the software engineers and system architects

communicate or interview domain experts and users

to gather requirements and their expectations. Based

on these, software engineers abstract requirements

into different models and combine them into a simple

system structure. This will then be presented to stake-

holders to illustrate how it meets their needs. How-

ever, due to the diverse backgrounds, there is, in fact,

no uniﬁed language or notations between these stake-

holders and developers (Khalajzadeh et al., 2020a).

It is more likely that some of the stakeholders can-

not understand what each model means, leading not

to reach a consensus on the software design. Mean-

while, development teams have to spend more time

explaining their progress and thoughts.

2.1 Example e-Health App Development

Consider a situation where a community is planning

to provide their residents with a novel e-Health mo-

bile application, which enables them to access med-

ical services more efﬁciently without the limitation

of time and location. The main objectives of this e-

Health application are to provide information regard-

ing common diseases, and to provide comprehensive

medical services such as medical consultation, pa-

tient registration, care plan generation, GP appoint-

ment booking, etc. Due to limited technical support,

the community decide to employ a technical team of

software engineers and system architects to develop

this application.

In this scenario the ﬁrst challenge would arise

when the community ﬁnds that the team lacked ex-

pertise in medical and health areas. This may lead

to inviting some medial experts to help the team

deal with medical-related issues. However, these two

teams would struggle, in communication as they lack

a uniﬁed language to share their ideas. For example

when the developers need to demonstrate their up-

dates, they won’t be able to do it with a code or a

program. They can switch to using a visual languages

like diagrams, but it is likely that the medical sup-

port team would still struggle in understanding these.

This would be mostly because the notations used in

these diagrams vary signiﬁcantly from the speciﬁc no-

tations used by the support team in their domain. This

makes the procedure of exchanging information take

a lot of time. It also reduces the contribution and in-

volvement of domain experts, as it is hard for devel-

opers to extract key information from them. There-

fore, as the complexity of the project grows, it would

become harder for stakeholders to track progress and

the collaborations would become difﬁcult to manage.

2.2 Key Challenges

As mentioned above, we explore two key challenges

of existing e-Health projects and these are:

• As the participation of experts from other ﬁelds

increases, more effective methods are needed to

involve experts, as some have no IT background.

• Existing visual languages cannot demonstrate e-

Health architecture well, since they have no uni-

ﬁed visual notations for e-Health concepts.

To solve these challenges, developers need a novel

uniﬁed language to illustrate their programs and to

better communicate with diverse stakeholders.

3 OUR APPROACH

The current e-Health area lacks a unifying language,

which makes communication between the develop-

ment team and domain experts difﬁcult and ambigu-

ous. Thus we adopted MDA to assist with the de-

velopment process as well as engage domain experts

better in the projects. In our approach, we introduce

a novel visual modeling language to model e-Health

applications. In this study, we ﬁrst conducted a sur-

vey to gather end-user requirements of e-Health and

to extract key concepts and components to include in

the meta-model. The preferences regarding the visual

ENASE 2024 - 19th International Conference on Evaluation of Novel Approaches to Software Engineering

284

notations of modeling tools from stakeholders in soft-

ware engineering projects were also collected. Based

on the ﬁndings, we selected various visual notations

for each of these concepts to design our Domain-

speciﬁc Modeling Language (DSML). Finally, with

SiriusWeb (an open-source low-code platform to de-

ﬁne custom DMSLs), we built a web-based modeling

environment to implement our modeling tool.

3.1 Requirement Collection

Currently there are visual modeling languages for

software development such as System Modeling Lan-

guage (SysML) and Uniﬁed Modeling Language

(UML). However, these require users to possess a cer-

tain level of IT and coding knowledge to be able to

use them. To design the most appropriate DSML for

e-Health, we needed a clear understanding of the core

components and concepts of e-Health and modeling

tools. Thus, we conducted a survey with different

human-centric aspects to gather user requirements of

e-Health applications as well as preferences for visu-

alizing modeling languages.

Survey Design. We designed an online survey with

around 30 questions and separated the survey into

three main sections, personal information, e-Health

requirements, and modeling tool preferences. To ben-

eﬁt from further exploring whether the preference and

requirement would be impacted by human character-

istics like gender, age, and role, we collected back-

ground information about all participants in the ﬁrst

section. The third section, modeling tool preferences,

focuses on the end users’ preference for the visual no-

tations of modeling tools, in order to build a more

accessible modeling tool for our end-users. Detailed

survey content can be accessed on (Shen, 2023).

Data Collection. We prepared questionnaires in En-

glish and used Google Forms to share them. We re-

ceived approval for this survey from the University

ethics committee. The survey was advertised via in-

vitation advertisement sent to researchers’ networks,

the University’s alumni network, social media posts,

and personal contact.

Data Analysis. We applied various descriptive

statistics to investigate the association between key

social-demographic factors, such as age, gender, ma-

jor, role, and others, and design elements. Partici-

pants’ choices of design elements for building their

own e-Health application were analyzed by the Chi-

square test, as such variables are all categorical and

can be converted to frequencies. For those partic-

ipants who had experience in visual-based system

modeling (12 of 20, 60%), preference for visual no-

tations and functions of modeling tools were also in-

Figure 1: Color preferences of different age groups.

vestigated using the Pearson Correlation test.

3.2 Requirements Analysis

Demographic Results. A total of 20 valid responses

were received to the online survey. Regarding demo-

graphic data, 13 men (65%) and 7 women (35%) par-

ticipated in our study. Participants ranged between

18 to 55 years old. Half of the participants (10 of 20,

50%) were between the ages of 18-25. Ninety percent

of the participants were living in Australia, but they

spoke a variety of languages, with only 35 percent

mainly speaking English at home. Furthermore, the

participants were predominantly students (80%) and

all majored in IT-related ﬁelds, namely Computer Sci-

ence, Information Technology, Artiﬁcial Intelligence,

Data Science, Software Engineering, and Cyber Se-

curity. Based on what they claimed, all participants

can be divided into 10 different role groups, such as

Software Developer, Computer Network Architect, IT

Project Manager, etc. Plus, 13 (65%) participants

responded that they had experience in using visual-

based modeling languages, and they were all willing

to share their preferences and experience with model-

ing tools with us.

e-Health Requirement. We ﬁrst asked participants

for the major reason for accessing the e-Health plat-

forms. A majority responded that they used it to ﬁnd

a General Practitioner (GP), while some who men-

tioned they used e-Health to track their health status.

We then provided participants with the user interface

of several e-Health platforms that are commonly used,

such as HotDoc and HealthEngine and asked them to

select their preferred one. The main reason for their

choices was the higher accessibility of the layout and

design elements on the user interface of these plat-

forms. Therefore, a more accessible and usable user

interface should be one of the key components of the

e-Health platform.

We assumed that people with different human as-

pects may have various preferences on the design. We

Human-Centered e-Health Development: An Accessible Visual Modeling Tool

285

found a signiﬁcant relationship between age and de-

sign preferences. A chi-square test on age and color

revealed that increasing age was signiﬁcantly corre-

lated with color preference for cool and warm color

themes (p = 0.05), indicating that cool color prefer-

ence showed a downward trend with age. However,

no signiﬁcant differences were found in preference

for icon design or text size with age. Older age groups

had a lower preference for cool tones, while those

aged 18 to 25 were more likely to prefer it (see Fig 1).

Our analysis of other human-centered factors showed

that there was no signiﬁcant relationship between de-

sign elements and these human factors.

Preferences of Modeling Tool. MDA emphasizes

the use of models during the development phase to

provide a comprehensive vision for the development

of software and system. We ﬁrst explored the partic-

ipants’ choice of notation styles. Respondents who

agreed to continue investigating the tool visual nota-

tion section were equally divided between icon-based

(38%) and shape-based (31%) symbol styles. A chi-

square analysis proved that age, gender and area were

indeed less relevant with notation style (p = 1). In

summary, diverse people have different preferences

on using visual notations for their e-Health applica-

tion diagrams, which indicates that providing multi-

ple visual notation styles are necessary to meet vari-

ous user requirements.

In this part of the survey, the importance of func-

tions and features in modeling tools were also rated

by participants. A Pearson correlation implemented

to ﬁnd the relationship between participants education

and their chosen tool function showed that there was

no statistically signiﬁcant relationship between these.

But over 75% participants thought that a recognizable

icon style is very necessary, as it is the most straight-

forward way to differentiate notations.

3.3 e-Health Meta-Model

There is currently no common language in e-Health,

making it difﬁcult for developers and stakeholders to

clearly communicate designs and share perspectives.

Therefore, we present a DSML to support e-Health

projects. DSML is deﬁned by meta-model, which rep-

resents important components in particular domains

and enables users to visualize and simulate system ar-

chitecture. Based on the ﬁndings of the survey, the

key entities of the e-Health meta-model are:

Application - represents the e-Health application it-

self. The application contains various entities to im-

plement the medical service and user interface. User

- represents all kinds of e-Health application users.

According to (Hesse and Shneiderman, 2007), there

Figure 2: e-Health meta-model.

Figure 3: Icon-based visual notations.

are 6 kinds of end users for e-Health, from healthcare

providers to individual patients. Health Record - rep-

resents types of health records, which bring together

various health information and physical examination

results from users, healthcare givers and medical or-

ganizations. UI - represents the user interface style

of the e-Health application, including theme color,

icon style, and so on, which enables end users to cus-

tomize the UI as they wish. Module - We deﬁned an

entity called a Module to encapsulate multiple func-

tions. Each Module has its unique name and feature,

but there can be multiple duplicated functions inside

it. Function - is the speciﬁc function embedded in

a certain module. A function can be repeated and

can also be different in different modules. Connector

- represents the relationships between entities within

the applications.

Multiple properties inside of each entity make

them more feasible and expressive. For example, UI

entity has design element properties that determine

how the user interface and other functions are pre-

sented to end users and how users can access medical

services more efﬁciently.

ENASE 2024 - 19th International Conference on Evaluation of Novel Approaches to Software Engineering

286

Figure 4: Shape-based visual notations.

3.4 e-Health Visual Notations

The new e-Health modeling language illustrates an e-

Health platform architecture at a broad and intuitive

level. There are no speciﬁc IT knowledge require-

ments to better support communication and collab-

oration between interdisciplinary team members to

identify key aspects and components of an e-Health

application such as its functional requirements, and

design elements. The ﬁndings from the requirement

survey indicated that most participants had a similar

tendency for two (shapes and icons) of three symbolic

styles that we provided in the survey, so we could not

directly conclude the most popular one. Therefore,

we presented two kinds of visual notations, domi-

nated by special icons and colored shapes, allowing

users to customise their preferred diagram style and

reducing the hurdle of involvement.

Icon-Based Notations. According to WCAG (An-

drew et al., 2018), to make visual elements perceiv-

able, our visual notation set comprises six distin-

guished special icons representing the core compo-

nents of e-Health platforms. The user icon (Fig 3 a)

indicates various user groups of e-Health, e.g. indi-

vidual patient, healthcare giver. The art icon (Fig 3

b) is intended to emphasize the design elements of

the user interface on the e-Health. Modules (Fig 3 f)

are presented by an icon constructed of several black

rectangles, meaning that it is composed of multiple

functions together, while its child functions are shown

by a gear icon (Fig 3 c) to emphasize their function-

ality. Web icon (Fig 3 e) is used to specify Appli-

cation models and health records are represented by

ﬁle icons (Fig 3 d). Finally, all individual component

models used in the same e-Health project need to be

connected to others by connectors (Fig 3 g). All icons

come with labels to improve their readability.

Shape-Based Notations. We assigned six distin-

guished shapes that are easy to differentiate, e.g. el-

lipse, rhombus, to present the shape-based visual no-

tations of our e-Health modeling tool (Fig 4). In order

to improve accessibility and provide a better visual

modelling experience, apart from the shape, we also

chose colors as the second feature to make it appro-

priate for all readers and users, including those who

are colorblind. Wong’s research project (Wong, 2011)

provides a palette of eight colors with good overall

variability for a visual design suitable for colorblind

people. Some of the colors in this palette are used in

our project to optimize our visual modeling symbols.

3.5 Graphical Editor

An experimental modeling tool was developed using

Eclipse SiriusWeb (https://www.eclipse.org/sirius/) to

support our DSML and corresponding visual nota-

tion sets. SiriusWeb is a framework to deﬁne cus-

tom web applications supporting modeling languages.

We chose SiriusWeb as custom models can be easily

shared with others by URL, so interdisciplinary team

members can collaboratively work on the modeling

projects simultaneously.

3.6 Speciﬁcation Generator

As our tool considers humans as the central part of

the modeling, it allows users to specify their features.

For instance, as shown in Fig 5, users can deﬁne the

UI style of the e-Health application with red color and

ﬂat icon style by using SiriusWeb graphic interface.

They can also specify their functions, e.g., booking

appointments. Then, our speciﬁcation generator is

able to export the models to ﬁles in text-based for-

mats like XML (Fig 6), which can be shared for re-

mote collaboration. Because XML is extensible and

ﬂexible, developers can use the XML ﬁle to deﬁne the

UI layout of e-Health apps based on the users’ needs.

Plus, Sirius enables generating code using Acceleo, a

template-based coding language that authorizes users

to create custom code generators. Since Sirius is built

on top of EMF, Acceleo can be utilized to generate

back-end code for any e-Health model in the EMF

format. Since we have not checked the collaboration

effectiveness between SiriusWeb and Sirius, custom

code generator will be our future plan.

4 EVALUATION

To determine the usability of our solution, we ﬁrst

tested our modeling tool with PoN (Moody, 2009)

and WCAG 2.1 (Andrew et al., 2018). Since PoN

deﬁnes the detailed design principles of visual sym-

bols, we can check our tool step-by-step and revise

it efﬁciently even in the absence of our experimental

subjects. Secondly, we organized a group user study

Human-Centered e-Health Development: An Accessible Visual Modeling Tool

287

Figure 5: Sample of e-Health models.

Figure 6: Sample code of e-Health model speciﬁcations.

to test the effectiveness of the modeling tool in the

real-world e-Health development environment.

4.1 Physics of Notations

PoN (Moody, 2009) is used as a guide to test the ef-

fectiveness of our visual notations. PoN-based visual

language assessment and improvement has been used

in many studies, for example, (Famelis and Chechik,

2019; Khalajzadeh et al., 2020b). Due to space limita-

tions, only the test results are presented in this section.

Semiotic Clarity: In our modeling tool, all entities

have 1:1 correspondence to their visual notations.

Perceptual Discriminability: Both our shape-based

notations or icon-based notations are differentiated by

various symbols, color and labels. Semantic Trans-

parency: In the newly built modeling tool, especially

in the icon-based view, special icons are used to rep-

resent models to minimize confusion. Complexity

Management: We minimize the visual elements as

much as possible to deal with the visual complex-

ity, for example, in each visual representation, there

is only one notation for each kind of model. Cogni-

tive Integration: Our project contains only one kind

of chart, thus this principle does not work in this case.

Visual Expressiveness: Shape-based view utilizes po-

sition, shape, texture, and color to distinguish sym-

bols and convey meaning, while icon-based only uses

shape and position. Dual Coding: All our visual sym-

bols have corresponding textual annotations to rein-

force and clarify meaning. Graphic Economy: As

the meta-model of e-Health contains only six entities,

its corresponding views do not include more symbol

types (6 model notations and 1 connector notation).

Cognitive Fit: Based on the results of the needs sur-

vey, our tool provides users with two sets of visual

symbols to meet their needs. However, as our tool

currently only supports to design and understand e-

Health system architecture, no more visual dialects

are available for other tasks.

4.2 User Study

We recruited 11 participants (seven Master/PhD stu-

dents, one academic staff, and three participants from

industry) within Faculty of IT at the University as

well as social media like LinkedIn and WeChat. All

participants had a certain level of IT background as

well as software engineering experience. 3 partici-

pants recognized themselves as software engineers, 1

as system architect, 3 as domain expert/business ana-

lyst/business manager, and 3 as data scientists. We

had 7 men and 4 women in the study. The study

was conducted through a one-hour online session via

Zoom. Participants were asked to complete a relevant

questionnaire, to collect their feedback.

A case study regarding e-Health application de-

velopment was ﬁrst provided with participants, and

we then asked them to choose any existing modeling

language like UML, BPMN, ER diagram, and mind

map, or their own ad hoc notations to model and de-

scribe their design on the canvas as the initial dia-

gram. We then introduced the concepts and notation

of the new modeling tool and asked them to model

the same e-health on a given SiriusWeb application

using our tool. Participants were required to choose

one of the visual notation sets and respond why they

prefer the chosen one. We randomly distributed the

newly drawn and initial diagrams among the partic-

ipants, while making sure that no one received their

own ones. In the next step, we asked participants to

compare the received diagrams and decide which one

they consider more proper to illustrate to other stake-

holders. Finally, they were asked to rate how easy the

new modeling tool was to understand, the IT knowl-

edge required to use it, and so on. Participants chose

a variety of visual modeling languages for their initial

diagrams, such as ER diagram and UML. The number

of people who chose ﬂowchart notations was the most

(n=4), while only one chose to use ad hoc notations.

In terms of participants preference of the two

ENASE 2024 - 19th International Conference on Evaluation of Novel Approaches to Software Engineering

288

Figure 7: Distribution of participants rating in terms of the

learning difﬁculty of the new modeling tool.

given visual notation sets (icon-based visual notations

and shape-based visual notations), seven responded

that they prefer special icons because they are simi-

lar to icons that they already knew, thus they can un-

derstand and recognize them more easily. Those who

preferred shape-based visual notations indicated that

compared with icons, shapes and colors help them to

differentiate various entities with less effort.

As shown in Fig 7, 10/11 participants mentioned

that the new modeling tool was easy to learn. The

participant related the reason for the neutral response

to the performance of SiriusWeb. Since all partici-

pants were invited to the online session at the same

time, the excessive trafﬁc volume led to a lag in the

response from the server, which decreased the learn-

ing experience of the participant during the user study.

The participant also indicated that it was a little difﬁ-

cult to learn how to use SiriusWeb as a novice, while

the modeling language and overall methodology were

good and easy to catch. All participants agreed that

using the new modeling tool did not require a high

level of IT knowledge and coding skills.

Overall, user feedback indicates that the new mod-

eling tool is very easy to learn and understand. Most

participants felt they can use the tool to easily com-

municate their ideas, techniques, and progress to

other stakeholders in a common language. However,

three participants still mentioned issues regarding the

poor responsiveness of the server and lack of cus-

tomization options. As the experimental proof-of-

concept tool was built on SiriusWeb, it helped us im-

plement the tool and provide it to the users, but it does

lack sufﬁcient functionality and good performance.

5 THREATS TO VALIDITY

Internal Threats. The ﬁrst threat to the validity

of our study is that there are currently no other suit-

able base modeling languages for modeling e-Health

software. We could not compare our approach with

any existing tools, therefore, we asked participants

to choose a modeling language of their own. How-

ever, most of the languages that they chose (such as

ﬂowcharts) were too general to be compared with our

modeling language effectively. Although it is not

an effective comparative experiment, the results still

helped us better understand users’ choice and percep-

tion of graphics and symbols.

External Threats. One of the external threats to

validity is the small number of participants. Another

issue is the lack of diversity of the people involved in

our program. All of our participants so far have come

from the IT ﬁeld. In the future we plan to conduct

user studies with health researchers to evaluate our

work more.

6 RELATED WORK

In this section, we summarize research regarding the

use of MDA for e-Health. The researchers applied

a novel approach called ActionGUI (de Dios et al.,

2014) for model-driven development of safety ap-

plications such as e-health. As e-health records are

highly sensitive, access to them must be controlled.

Extensions to SecureUML are used to specify secu-

rity policies for data models in this approach. Se-

cureUML extends role-based access control (RBAC)

with authorization constraints, which supports the

modeling of roles and other constraints. This pa-

per also proposes a complete meta-model for an e-

Health application containing 18 entities, such as pro-

fessionals, medical centers, doctors, etc. The Ac-

tionGUI code generator can then automatically gen-

erate a ready-to-deploy secure e-Health application.

In (Kotronis et al., 2018), researchers applied

SysML to explore the application of a model-driven

approach to the Internet of Things (IoT) e-Health sys-

tems and highlighted key requirements. Through e-

Health case studies, namely Remote Elderly Monitor-

ing Systems (REMS) and Smart Ambulance System

(SAS), the identiﬁcation of criticality in healthcare

IoT systems is explored as a ﬁrst step to effectively

manage them in system implementation and deploy-

ment. They then used SysML to model a range of

REMS structures. As a result, they found that reason-

able system-wide abstractions are helpful to systems

engineers as they can clarify the design and manage-

Human-Centered e-Health Development: An Accessible Visual Modeling Tool

289

ment of complex mixed-critical e-Health systems.

In summary, the existing works demonstrated the

feasibility and efﬁciency of MDA in real work scenar-

ios. MDA is able to lower the threshold for develop-

ers to understand project requirements by abstracting

various components of the system. However, none of

these works can model speciﬁc user’s preferences for

e-Health applications such as the website appearance

and the functionalities they prefer. Also, they have

not developed a solution for medical experts and de-

velopers to work collaboratively. Although existing

approaches have improved the efﬁciency of software

development, they do not cater for the participation

of experts in app development, making them unable

to give professional insights, which results in the ﬁnal

deliverable not being able to satisfy the users.

7 CONCLUSION

We have presented a novel visual-based e-Health

modeling language. The goal of our DSML is to en-

able people from diverse ﬁelds to engage in e-Health

design projects better. It describes core concepts and

components of e-Health. Based on the ﬁndings from

the user requirements survey, we provided our users

with two accessible and distinguishable sets of visual

notations, depending on the Web Content Accessibil-

ity Guidelines 2.1 (Andrew et al., 2018), for designing

the visual diagram describing the custom e-Health ap-

plication. We evaluated the accessibility and usability

of our tool through a Physics of Notations assessment

and a group end user study. Evaluation results illus-

trate that all participants responded well in terms of

the usability of both language and overall approach.

ACKNOWLEDGMENT

Madugalla is supported by ARC Laureate Fellowship

FL190100035.

REFERENCES

Andrew, K., Joshue, C., Alastair, C., Michael, C., Cald-

well, B., Cooper, M., Reid, L. G., Vanderheiden, G.,

Chisholm, W., Slatin, J., and White, J. (2018). Web

content accessibility guidelines (wcag) 2.1. WWW

Consortium (W3C).

Brambilla, M., Cabot, J., and Wimmer, M. (2017). Model-

driven software engineering in practice. Synthesis lec-

tures on software engineering, 3(1):1–207.

Ceri, S., Fraternali, P., and Bongio, A. (2000). Web mod-

eling language (webml): a modeling language for de-

signing web sites. Computer Networks, 33(1-6):137–

157.

de Dios, M. A. G., Dania, C., Basin, D., and Clavel, M.

(2014). Model-driven development of a secure ehealth

application. Engineering Secure Future Internet Ser-

vices and Systems: Current Research, pages 97–118.

Famelis, M. and Chechik, M. (2019). Managing design-

time uncertainty. Software & Systems Modeling,

18:1249–1284.

Fischer, K., Krumeich, J., Panﬁlenko, D., Born, M., and

Desfray, P. (2014). Based modeling: A stakeholder-

centered approach for model-driven engineering. In

Advances and applications in model-driven engineer-

ing, pages 317–341. IGI Global.

Hause, M. et al. (2006). The sysml modelling language. In

Fifteenth European Systems Engineering Conference,

volume 9, pages 1–12.

Hesse, B. W. and Shneiderman, B. (2007). ehealth research

from the user’s perspective. American journal of pre-

ventive medicine, 32(5):S97–S103.

Khalajzadeh, H., Simmons, A. J., Abdelrazek, M., Grundy,

J., Hosking, J., and He, Q. (2020a). An end-to-

end model-based approach to support big data ana-

lytics development. Journal of Computer Languages,

58:100964.

Khalajzadeh, H., Verma, T., Simmons, A. J., Grundy, J.,

Abdelrazek, M., and Hosking, J. (2020b). User-

centred tooling for modelling of big data applica-

tions. In 23rd ACM/IEEE International Conference on

Model Driven Engineering Languages and Systems:

Companion Proceedings, pages 1–5.

Kotronis, C., Nikolaidou, M., Dimitrakopoulos, G., Anag-

nostopoulos, D., Amira, A., and Bensaali, F. (2018).

A model-based approach for managing criticality re-

quirements in e-health iot systems. In 2018 13th

annual conference on system of systems engineering

(SoSE), pages 60–67. IEEE.

Moody, D. (2009). The “physics” of notations: toward a sci-

entiﬁc basis for constructing visual notations in soft-

ware engineering. IEEE Transactions on software en-

gineering, 35(6):756–779.

Shen, J. (2023). Human-centred web-based modelling tool

data.

WHO (2019). WHO guideline: recommendations on digital

interventions for health system strengthening: execu-

tive summary. Technical report, World Health Orga-

nization.

Wong, B. (2011). Color blindness. nature methods,

8(6):441.

Zhuang, W., Gan, X., Wen, Y., and Zhang, S. (2022).

Easyﬂ: A low-code federated learning platform

for dummies. IEEE Internet of Things Journal,

9(15):13740–13754.

ENASE 2024 - 19th International Conference on Evaluation of Novel Approaches to Software Engineering

290