International Journal of Medical Informatics 167 (2022) 104860

Contents lists available at ScienceDirect

International Journal of Medical Informatics

journal homepage: www.elsevier.com/locate/ijmedinf

Medical informatics and digital health multilingual ontology (MIMO): A

tool to improve international collaborations
Arriel Benis a, b, h, *, Julien Grosjean c, d, Kévin Billey c, Gustavo Montanha e, Verena Dornauer f,
Mihaela Crișan-Vida g, h, Werner O Hackl f, Lăcrămioara Stoicu-Tivadar g, h, Stéfan
J. Darmoni c, d, h
a
Faculty of Industrial Engineering and Technology Management, Holon Institute of Technology, Israel
b
Faculty of Digital Technologies in Medicine, Holon Institute of Technology, Israel
c
Department of Biomedical Informatics, Rouen University Hospital, France
d
LIMICS Laboratory of Medical Informatics and Knowledge Engineering in e-Health, INSERM U1142, Sorbonne Université, France
e
Centro Hospitalar de Entre o Douro e Vouga EPE: Santa Maria da Feira, Aveiro, Portugal
f
Institute of Medical Informatics, Private University for Health Sciences, Medical Informatics and Technology, Hall in Tirol, Austria
g
Department of Automation and Applied Informatics, University Politehnica Timișoara, Timișoara, Romania
h
European Federation for Medical Informatics, Le Mont-sur-Lausanne, Switzerland

A R T I C L E I N F O A B S T R A C T

Keywords: Background: Even if English is the leading language for international communication, it is essential to keep in
Vocabulary, Controlled mind that research runs at the local level by local teams generally communicating in their local/national lan­
Communication guage, especially in Europe among European projects.
Information Services
Objective: Therefore, the European Federation for Medical Informatics - Working Group on Health Informatics for
Language
Digital Health
Inter-regional Cooperation” has one objective: To develop a multilingual ontology focusing on Health Informatics
Translations and Digital Health as a collaboration tool that improves international and, in particular, European collaborations.
Biomedical Ontologies Results: We have developed the Medical Informatics and Digital Health Multilingual Ontology (MIMO). Hosted on
Intersectoral Collaboration the Health Terminology/Ontology Portal (HeTOP), MIMO contains around 1,000 concepts, 460 MeSH De­
scriptors, 220 MeSH Concepts, and more than 300 newly created concepts. MIMO is continuously updated to
comprise as recent as possible concepts and their translations in more than 30 languages. Moreover, the MIMO’s
development team constantly improves MIMO content and supporting information. Thus, during workshop
discussions and one-on-one exchanges, the MIMO team has collected domain experts’ opinions about the com­
munity’s interests and suggestions for future enhancements. Moreover, MIMO will be integrated to support the
annotation and categorization of research products into the HosmartAI European project involving more than 20
countries around Europe and worldwide.
Conclusion: MIMO is hosted by HeTOP (Health Terminology/Ontology Portal), which integrates 100 terminol­
ogies and ontologies in 55 languages. MIMO is freely available online. MIMO is portable to other knowledge
platforms as part of MIMO’s main aims to facilitate communication between medical librarians, translators, and
researchers as well as to support students’ self-learning.

Abbreviations: API, Application Programming Interface; DSM, Diagnostic and Statistical Manual of Mental Disorders; EFMI, European Federation ofMedical
Informatics; EU, European Union; HeTOP, Health Terminology/Ontology Portal; HIIC, Health Informatics for Inter-regional Cooperation; HosmartAI, HOspital Smart
development based on Artificial Intelligence; ICD, International Classification of Diseases; ISO, International Organization for Standardization; MedDRA, Medical
Dictionary of Regulatory Activities; MeSH, Medical Subject Headings; MIE, Medical Informatics Europe Conference; MIMO, Medical Informatics and Digital Health
Multilingual Ontology; NCIt, National Cancer Institute Thesaurus; OMT, Online Machine Translation; OWL, Web Ontology Language; SNOMED CT, Systematized
Nomenclature of Medicine - Clinical Terms; STC, Special Topic Conference; UMLS, Unified Medical Language System; WG, Working Group.
* Corresponding author at: Faculty of Industrial Engineering and Technology Management, Holon Institute of Technology, Golomb St. 52, Holon, Israel
E-mail address: arrielb@hit.ac.il (A. Benis).

https://doi.org/10.1016/j.ijmedinf.2022.104860
Received 24 March 2022; Received in revised form 10 May 2022; Accepted 24 August 2022
Available online 29 August 2022
1386-5056/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
A. Benis et al.
1. Introduction
1.1. Background
Scientific projects, particularly in the Health Sciences and Technol­
ogies field, are nowadays increasingly multi-centric by involving part­
ners with different capabilities, skills, and expertise from different
countries [1,2]. This complexity always happens in European projects
involving other partners using at least one European language, even if
English became the scientific “lingua franca” used for international
communication. Indeed, English is massively used compared to other
languages for scientific publications. This is not due to the English
fluency of the authors owing to different research evaluation processes
highlighting the importance of publishing in English-speaking interna­
tional journals [3]. However, it is essential to keep in mind that research
runs at the local level by local teams generally communicating in their
own local or national language [4,5].
Digital Health and Health Informatics are at the crossroads of med­
icine and health sciences, computer science and engineering, informa­
tion and communication sciences, mathematics, statistics, technology,
and innovation management. However, Digital Health focuses on using
information and communication technologies to deliver healthcare
services to promote better health and well-being. Health informatics
deals broadly with data, information, knowledge, and decision-making
for healthcare professionals and organizations. In both fields, the in­
teractions between different kinds of actors, such as healthcare infor­
maticists, practitioners, managers, and customers and their relatives, are
critical and complex. However, the main objective of Digital Health is to
support the delivery of healthcare in a safe, efficient, and accessible way
to and for everyone.
The European Federation for Medical Informatics (EFMI) is built
around 30 member-countries affiliated societies [6]. One of the objec­
tives of EFMI is to encourage academic and professional collaboration
and cooperation between all its members (both from academia and in­
dustry), to enhance health and medical informatics research, education,
knowledge, and industrial transfer. Thus, the EFMI Working Group
(WG) on Health Informatics for Inter-regional Cooperation (HIIC) [7]
deals with enhancing the quality of the exchanges between international
health informatics research partners.
1.2. Aims and objectives
In the 1990s, some glossaries, thesauri, and domain models,
comprising a few hundred to 2,000, were developed in Medical Infor­
matics. Two teams led these projects, one from Missouri, USA [8,9] and
the other from Liverpool, UK [10]. These paper-printed vocabularies
were in English only. Moreover, according to our knowledge, there is no
similar prior work dealing with multilingual issues. One of the EFMI WG
HIIC aims to develop a frequently updated multilingual controlled vo­
cabulary focusing on Health Informatics and Digital Health as a
collaboration tool that improves international and, more particularly,
European cooperation [11]. Thus, the leading question of this research
relates to the definition of a methodology allowing the continuous
development and improvement of the multilingual ontology of medical
informatics.
In this manuscript, our objective is to present the development
process of the multilingual ontology on health informatics (and beyond,
digital health, including health information sciences), called MIMO
(Medical Informatics Multilingual Ontology). The MIMO project mainly
focuses on European languages within the umbrella of EFMI. However,
it includes other languages worldwide to be accessible to the largest
possible part of the scientific community. Moreover, we show that
MIMO is favorably welcomed by the “Health Informatics and Digital
Health” community both at the individual and collective levels.
2
International Journal of Medical Informatics 167 (2022) 104860
2. Material and methods
The development, management, and continuous improvement of
MIMO are challenging. MIMO was built in five steps. The first version of
MIMO was developed in 2018 as a flat dictionary [12]. Said version was
a proof-of-concept based on the first three steps of the MIMO develop­
ment methodology reported below. Then the current version of MIMO
consisted of combining knowledge extraction (from existing controlled
vocabularies), and knowledge generation (by getting new terms from
domain experts, and from native speakers of the different included
languages in MIMO, translations of previously translated ones).
The first step of the MIMO project consisted in defining the list of the
official languages of the EFMI member countries.
In 2021, the 27 member countries of the European Union (EU) had a
population of 517 million inhabitants and 746 million, including the
people of geographical Europe and associated countries. Currently,
there are 24 official languages in the EU: Bulgarian, Croatian, Czech,
Danish, Dutch, English, Estonian, Finnish, French, German, Greek,
Hungarian, Irish, Italian, Latvian, Lithuanian, Maltese, Polish, Portu­
guese, Romanian, Slovak, Slovene, Spanish and Swedish, and 36 lan­
guages in Europe (e.g., Russian) [13,14]. According to their origin and
similarities, all these languages can be classified into eight language
groups [15].
The first version of MIMO was a flat dictionary as the third step with
some initial main categories of terms such as “Computer (Hardware),”
“Computer (Software),” “Medical device,” “System,” “Programming,”
“Organization,” “Journal,” and “Application field.” An initial set of 160
terms populated the dictionary, and when relevant related acronyms
existed for some terms, they were also added. For example, “Interna­
tional Conference on Informatics, Management, and Technology in
Healthcare” and “ICIMTH,” in English, French, Romanian, and Hebrew,
the main languages known as a native-speaker level by the EFMI dedi­
cated team to this project.
The second step consisted in associating each of the 36 languages
that have been previously identified with the relevant ISO 639 code to
facilitate the subsequent implementation steps [16].
Then, as a third step, a simple automated translation process, using
an online machine translation (OMT) tool supporting over 100 lan­
guages at various levels, has been implemented. We listed the terms and
used an OMT [17] function for each of the 36 targeted languages. The
OMT service was able to detect the language of the submitted term with
high accuracy and then translate it with variable correctness to target
languages according to the ISO 639 codes previously selected. As a result
of this automated translation process, a flat file has been generated.
However, as a part of the quality control point of this step, the MIMO
team has pointed out the lack of translated accuracy for at least half of
the 36 languages. This lack of accuracy was noticed by having, for
example, English terms used in the translation to another language
instead of said language itself or for languages known by the team in­
accuracy (e.g., informatics translate in Hebrew as “Library Science”
instead of “Information Science”). This issue is due to the different levels
of maturity of the translation capabilities of the service. These results are
consistent with other studies focusing on other fields [18] and specif­
ically in health [19]. Accordingly, human language speakers are
involved in validating, improving, and expanding the dictionary over
time.
The fourth step to create MIMO was the creation of a thesaurus to be
used to build an ontology. The MIMO team has analyzed the main
controlled vocabularies in health and being a part of UMLS [20]: MeSH
[21], SNOMED CT [22], NCIT [23], MedDRA [24] or ICD (in their ninth,
10th, and 11th revision) [25]. Based on this manual review by two
medical informaticians (AB & SJD), the MeSH thesaurus was assessed as
having the most comprehensive number of medical informatics con­
cepts. Indeed, the MeSH thesaurus is the unique reference terminology,
where a tree describing medical informatics and, more broadly, infor­
mation science exists. Accordingly, to complete this thesaurus as
A. Benis et al.
exhaustively as possible, the MIMO team has decided to expand the
concepts to be included not only extracted from the MeSH hierarchy
“medical informatics” but to be extracted from the broader concept
“information science.” In other words, there is an “IS_A” relation be­
tween “medical informatics” and “information science,” where “medical
informatics” is the narrower concept. No such tree exists in SNOMED CT
[22] or in NCIT [23], where only a tree of “software” exists in these two
reference terminologies, and both are currently integrated into MIMO.
The MeSH thesaurus has five different types of concepts: MeSH De­
scriptors, MeSH Supplementary Concepts, MeSH Concepts, MeSH
Resource Types, and MeSH qualifiers. Two of which were used to build
MIMO: MeSH Descriptors and MeSH Concepts of the current 2021
version. Terms in a MeSH record, which are strictly synonymous with
each other, are grouped in a category called a “MeSH Concept.” Each
MeSH record consists of one or more concepts, and each concept consists
of one or more synonymous terms. Each MeSH concept is related to a
MeSH Descriptor. For example, the MeSH Descriptor “Diagnostic and
statistical manual of mental disorders” has four related MeSH Concepts:
DSM-II, DSM-III, DSM-IV, and DSM-V [26]. These four MeSH concepts
were integrated into MIMO. Four relations exist between MeSH De­
scriptors and MeSH Concepts [21]:
1. Preferred Term (PT), the MeSH Descriptor, and the MeSH Concept
are accurate synonyms;
2. Broader Term (BT), the MeSH Concept is hierarchically more general
than the MeSH Descriptor;
3. Narrower Term (NT), the MeSH Concept is hierarchically more
limited than the MeSH Descriptor;
4. Related Term (RT), the MeSH Concept, and the MeSH Descriptor are
related, is a “see also” relation, without any hierarchical relation (e.
g., prison RT prisoners).
Thus, in the example of the MeSH Descriptor “Diagnostic and sta­
tistical manual of mental disorders,” MIMO has applied the relations
between this MeSH Descriptor and the four MeSH Concepts and were
integrated into MIMO, which is not the case in the MeSH thesaurus.
The previous concepts included in the dictionary were disseminated
in the various hierarchies of MIMO. Specific hierarchies were created
with medical concepts, which do not exist in other systems, methods,
and terminologies of reference: in particular, “Supervisory Control And
Data Acquisition (SCADA)” [27,28] or “diseases linked to digital health”
[29,30]. Some added concepts were coming from SNOMED CT, Med­
DRA, and NCIT.
The structure of the tree-based MeSH has implications for the
multilingual ontology by pointing out the exact or near translation of
terms, such as native speakers provide them (e.g., humans or
knowledge-sharing platforms such as Wikipedia). Nowadays, MIMO is
built as an ontology according to the different relationships and de­
pendencies defined in the hierarchy of terms. For example, translations
are defined as follows: is_a_human_translation, is_a_terminology-
based_translation.
In the fifth step, because disseminating the information and the
knowledge included in MIMO is a critical phase, the MIMO team had to
define which kind of platform would be relevant to:
• Host and maintain it efficiently over time,
• Support integration with other existing terminologies, classifica­
tions, and ontologies in health and engineering sciences,
• Provide the end-user with a user-friendly interface that will
encourage using the dictionary as a day-to-day tool.
The MIMO team’s interest focused on the following terminology
server: a cross-lingual health terminology and ontology server (HeTOP;
URL: https://www.hetop.eu/hetop/rep/en/terminologies/ [31]),
which contains more than 2 million concepts, 100 terminologies, and
ontologies, 55 languages, including non-European languages (e.g.,
3
International Journal of Medical Informatics 167 (2022) 104860
Mandarin, Arabic, Korean or Japanese) [32]. Moreover, the Generic
Model of HeTOP is compatible with the ISO 25,964 standard and allows
interoperability with other vocabularies platforms [33–35]. The proof-
of-concept of the MIMO was integrated, in mid-2019, into the HeTOP
cross-lingual terminology server and is available to the consortium using
an identification [36]. Furthermore, we have evaluated the interest of
MIMO by the “Health Informatics and Digital Health” community both
at the individual and collective levels.
3. Results
3.1. MIMO as a community freely available tool
MIMO is currently available as a component integrated into the
HeTOP server(URL: https://www.hetop.eu/hetop/rep/en/EFMIMIMO/
). Among the 100 terminologies and ontologies integrated into HeTOP,
only 18 are also integrated into the UMLS [37]. Termino-ontologies are
regularly updated and are accessible via a website and a dedicated web
service. HeTOP has been designed as a reference multi-terminology and
cross-lingual portal to help librarians, translators, students, and medical
professionals retrieve resources and knowledge across various complex
medical fields.
On 27 April 2022, MIMO contained 1,261 concepts, 666 from MeSH,
136 from NCIT, 177 from SNOMED CT, and 282 newly created concepts.
We plan to integrate around 2,500 MIMO concepts at the end of the
HosmarAI project in 2024 and continue to expand it in the future.
The advantages of using HeTOP as the hosting platform of MIMO are:
• The navigability capabilities between the different available the­
sauruses and ontologies allowing discovering if a search term exists
in MIMO and provided resources;
• The interface is translated into 12 languages (partially or fully pro­
cessed both by domain experts and online AI-based systems) to
facilitate the platform used by a large community not always
comfortable with English, similarly to other well-known
environments;
• The possibility to get dynamically generated links to relevant search
query results on PubMed, or at this time, CISMeF, the Catalog and
Index of French Language Health Resources on the Internet [38];
• The frequent updates of the terminologies and ontologies hosted
allowing to be enriched thanks to alignment processes MIMO, when
relevant dynamically;
• HeTOP comprises capabilities of controlled vocabularies alignment
and mapping [39,40]. Thus, the initial terms included in MIMO have
been automatically aligned with the same ones existing in other
terminologies and ontologies existing in HeTOP. Then the MIMO
terms were mapped in the vocabularies wherein its terms were
discovered. It allowed expanding MIMO automatically. Neverthe­
less, health informatics and digital health fields are evolving quickly.
Some terms included in MIMO are not discovered in other HeTOP
vocabularies. This is the most advanced added value of this research
(i.e., having an up-to-date vocabulary comprising terms related to
cutting-edge research and technologies). The alignment and map­
ping processes are periodically run to ensure the availability of the
most enriched version of MIMO;
• The possibility of involving a native language speaker and domain
expert in the translation process or validation to a specific language
without an open system.
Fig. 1 shows an example of a HeTOP response for a query with the
term “bibliometrics.” Using the interface in its English version (the user
can switch to a different language by using the dropdown list appearing
on the top-left side of the screen) and asking for results from all the
available terminologies and ontologies (by selecting in the tree of the
available terminologies “Terminologies selection”) allowing a matrix
navigation (among languages and terminologies). The following
A. Benis et al.
Fig. 1. Example of HeTOP response for a query with the term “bibliometrics”.
description sections are available:
• Preferred Label: the MeSH Descriptor and its translated versions as
provided by native speaker level domain experts; herein, e.g., “bib­
liometrics” is translated into 13 languages;
• We have developed a specific generic algorithm (currently named
WikiMeSH) to propose an extended translation of any thesaurus
based on Wikipedia pages [41]. A mapping was realized between
each MeSH descriptor (preferred terms and synonyms) and corre­
sponding Wikipedia pages by using the Wikipedia API (Application
Programming Interface) [42]. Among the top 20 languages of this
study, seven currently have no MeSH translations: Arabic, Catalan,
Farsi (Iran), Mandarin Chinese, Korean, Serbian, and Ukrainian. For
these seven languages, WikiMeSH is proposing translations, for
example, of 47 % to Arabic and 34 % to Serbian [43] ;
• MeSH synonyms: Synonyms and equivalent terms and concepts ac­
cording to the MeSH but less used or less popular or more generally
“related terms”; herein, we got “bibliometrics” MeSH synonyms in
British English, American English, and Danish;
• DeCS synonyms: based on MeSH, DeCS (in Spanish: Descriptores en
Ciencias de la Salud, in English: Health Sciences Descriptors) is a
controlled and structured vocabulary in four languages (Spanish,
Portuguese, French, and English). It expands the MeSH thesaurus,
containing over 5,000 health concepts. The DeCS project started
during the 1970 s [44]; herein, we got “bibliometrics” translated by
domain experts to Spanish and Portuguese;
• Définition CISMeF: definition of the preferred label, in French, when
available, manually translated by CISMeF librarians and health
professionals [38];
4
International Journal of Medical Informatics 167 (2022) 104860
• MeSH definition: definition of the preferred label, in English,
extracted from the MeSH [21];
• Wikipedia link: Links to the relevant and available Wikipedia pages
in the different languages ran into MIMO to provide the users with
more definitions and information as the Wikipedia community writes
and shares it. It is critical to point out that Wikipedia’s community
controls the content of Wikipedia, but this does not ensure that it is
validated by medical informatics and digital health vocabularies
experts.
The HeTOP user can search for a term only in MIMO. To do this, the
user must select only MIMO in the list “Terminologies selection”.
3.2. Health informatics and digital Health” community interest in MIMO
As a part of the development of MIMO and keeping in mind the need
for its continuous improvement, getting feedback is essential, both at the
individual and collective level. We received inputs broadly (1) during
workshop discussions and one-on-one exchanges and (2) in integrating
MIMO into the HosmartAI European project for supporting multilingual
knowledge and research products dissemination.
3.3. Medical informatics community feedback on the MIMO concept
Before starting the MIMO project in its current format, the EFMI WG
HIIC had organized, during scientific meetings, a few workshops and
panels focusing on cooperation and collaboration in the health com­
munity during international scientific meetings. During these events, we
have discussed with the audience the relevancy and our research on a
A. Benis et al.
dedicated multilingual controlled vocabulary (ICIMTH 2019 [12]; EFMI
and the Romanian Society of Medical Informatics - Adapting Education
during COVID-19 Pandemic (online event) [45]; MIE 2021 [46]; STC
2021/HosmartAI Satellite event [47]). We got encouraging comments
on the MIMO concept from the health informatics and digital health
community. Indeed, end-users shared an interest in integrating and
translating knowledge for the clinical practice and health services cus­
tomers’ benefits. Moreover, from a technological perspective, one
frequent feedback was that a multilingual ontology focusing on medical
informatics and digital health, such as MIMO, is fundamental to the
future of health information technology systems. It must support the
perpetual inclusion of new technological tools in the health-related
process (i.e., diagnosis, treatment, monitoring, and follow-up). More­
over, one crucial community comment is that MIMO should be open-
sourced like other knowledge management tools to allow its dissemi­
nation, growth, and maintenance. Therefore, MIMO is freely accessible
on HeTOP to the entire scientific community and on-demand as an OWL
(Web Ontology Language) file and via the HeTOP API.
Moreover, it was also important to have clinicians’ feedback about
potential applications to daily practice. On the one hand, we got these
non-formal, verbal, and unstructured inputs, from members of the
MIMO team (JGMMM, VD), and on the other hand, from a few partici­
pants at the dedicated meetings reported above. A large part of them are
looking at MIMO as the pillar of tools that allows them to find semantic
relations in their clinical notes and their knowledge base (e.g., scientific
papers, reports) to be able to support and justify some digital health-
oriented advice given to patients (e.g., using a smartwatch for moni­
toring physical activity, preferring a non-invasive, robot-assisted sur­
gery) as a part of an evidence-based medicine approach.
3.4. Integration with the “HOspital SMART development based on
artificial Intelligence” European project
Since January 2021, MIMO development has been partially granted
by the Horizon 2020 European project “HOspital SMART development
based on Artificial Intelligence” (AI; HosmartAI) [48]. HosmartAI is a
European project involving 24 partners in 12 countries. HosmartAI
promotes an effective and efficient healthcare system transformation
through technological developments in AI and robotics. HosmartAI is
creating a common open integration platform, called HosmartAI
Marketplace, with tools to facilitate and measure the benefits of inte­
grating digital technologies (such as robotics and AI) in the healthcare
system. The main objective herein is to improve the interoperability
among the various tools of this project. Accordingly, MIMO has been
selected for allowing the description with the MIMO concepts of the
different products available in the HosmartAI Marketplace. The main
objectives of this integration are to facilitate (1) the uses of the research
products developed by the 24 partners in 12 countries, and (2) the daily
communication practice of the different teams in one of their mother
tongues.
With this objective and the previous inclusion of the MeSH tree on
“information science,” we noticed that its granularity is limited (e.g., the
term “robotics” appears a few times in MeSH, but not the biomedical
engineering terms related to the components of a robot). Therefore, this
framework expands MIMO to provide a more exhaustive and up-to-date
controlled vocabulary relevant to hospital informatics.
Moreover, MIMO can be integrated with and into other platforms. To
this end, MIMO is accessible as an OWL file and a Web service via the
HeTOP API for the HosmartAI consortium to be integrated into the
platforms developed.
In the context of the HosmartAI project, to support the overall
communication of different actors of the projects and the users of its
products efficiently and effectively, overtime MIMO is:
• Expanding the number of concepts (from nearly-one thousand to a
few thousand);
5
International Journal of Medical Informatics 167 (2022) 104860
• Expanding the number of categories and sub-categories (to a few
dozen). For example, robotics, Internet-of-Things, and Human-
Computer Interactions, in the context of digital health and artifi­
cial intelligence in healthcare (i.e., hospitals) should be expanded;
• Involving native speakers of the different European languages for
evaluating, validating, or correcting the automatically imported or
translated concepts, as we did for Portuguese, Romanian, and soon
Swedish.
4. Discussion
4.1. Overview
The development of multilingual controlled vocabularies such as
thesauri and ontology related to the Medical Informatics and Digital
Health fields for enhancing the quality and efficiency of international
projects is challenging. The MIMO developers aim to support all the
actors interacting in multilingual projects at all the levels of professional
health (industry) communication (e.g., education, research, engineer­
ing, knowledge dissemination, marketing), such as the one required in
the European context. The MIMO benefits from the computational, sci­
entific, and linguistic resources offered by the freely accessible HeTOP
environment.
5. Strengths and limitations
The globalization process of the healthcare system, looking at the
clinical practitioners, administrative teams, technology developers, and
enablers, is long and complex. Indeed, the rules and regulations are
specific to each country. Moreover, the spoken language in each country
is specific, so the same word can have another meaning when translated.
The main strength of MIMO is the capacity to provide to the different
actors of a healthcare technologies-related project to communicate in a
more unified language framework. MIMO gives the possibility to search
in their native language a translation of medical informatics, or more
broadly, a digital health-related term currently in another European
language, and check that the meaning is the same for both in a third one,
like English.
The second strength of MIMO is the capability to combine automated
validated and domain-expert translations. For languages primarily
practiced in Europe and around the World, the automated and validated
translations from different sources cover a large part of the MIMO terms
and concepts. Nevertheless, native language speakers’ domain experts’
involvement allows getting an up-to-date terminology and thesaurus in a
range of different cases, such as:
1. New terms and concepts that are still not in existing terminologies
and ontologies, such as MeSH,
2. For less popular languages with a meager rate or non-existing
translations.
A third strength of MIMO is that HeTOP hosts it, and this allows:
1. Hosting and maintaining it efficiently over time;
2. Supporting the mapping with other existing terminologies, classifi­
cations, and ontologies in health and engineering sciences;
3. Providing to the end-user a user-friendly interface that will
encourage using MIMO as a day-to-day tool; each MIMO concept has
a dedicated URI that can be shared with end-users across the globe.
The main strength of MIMO is to be up-to-date with recent and new
terms and concepts related to new technologies.
However, MIMO has some limitations. Indeed, for non-frequent lan­
guages, a native speaker level domain expert with multilingual capabil­
ities is needed to translate from one language to another or validate
automated translation (e.g., based on online services such as Google
A. Benis et al.
Translate, Bing Translator). Thus, for this domain expert’s time-
consuming involvement, it is necessary to engage them in a community
also giving translation peer-review, validation, and enhancing domain-
related knowledge to expand the new terms’ translation capabilities.
Moreover, not all the terms and concepts of the MIMO are fully
translated into all the supported languages simultaneously. This time-
based discrepancy and human resources limitations are, like in other
multilingual projects (i.e., Wikipedia [41], Wiktionary [49]), well-
known issues that can be managed by building a community of volun­
teers worldwide.
Medical informatics and digital health are dynamic fields with
frequent new inputs (i.e., new technologies, systems, methods, algo­
rithms, new terms and concepts). Therefore, the MIMO core team and
community must have up-to-date knowledge and define a flow for
integrating new terms and concepts (e.g., which languages are the initial
inputs and how are they curated? What is the frequency of the MIMO
updates release). Furthermore, this is in continuation of the previous
point.
5.1 Future perspectives.
The next scheduled steps of the MIMO consist in:
• Developing a community of volunteer domain experts in the different
fields of medical informatics and digital health having multilingual
capabilities for expanding both the translations of the existing terms
and developing a cross-translators (humans and AI-based systems)
validation process;
• Expanding MIMO to other non-European languages, such as Arabic,
Chinese, and Russian, as part of the UN official languages in addition
to English, French, and Spanish [50];
• Developing tools allowing to capture existing translations in open
data resources and so expanding MIMO to other languages supported
by multilingual collaborative communities [43];
• Adding vocabularies of trending or emerging fields in medical
informatics and digital health, such as “One Digital Health” [51],
“Accident and Emergency Informatics” [52], robotics in healthcare
systems and industry [53];
• Moving MIMO to a complete multilingual ontology to facilitate its
maintenance and expansion to support efficient international health
technologies-related projects [54].
6. Conclusion
Developing persisting and fruitful inter-regional and international
projects, cooperation, and collaboration is an endless challenge. Indeed,
workspace virtualization, online-based communication, and the
continuously increasing use of computational tools facilitate the search
for new academic and industrial research partners. Nevertheless,
communication misunderstandings can lead to delays and failure of
potentially disruptive innovation projects [55,56]. Therefore, managing
complex projects in medical informatics and digital health fields re­
quires a basic but proper understanding of different relevant vocabu­
laries (e.g., technology, informatics, engineering, health, and medical
sciences) [57,58].
MIMO aims to provide a multilingual tool focusing specifically on
health informatics and digital health technologies. The methodology
empirically implemented for continuous development of MIMO strives
to have this controlled vocabulary as a pillar component of an envi­
ronment supporting health sciences information professionals’ different
activities, facilitating the mutual understanding of researchers both
from academia and industry, and enhancing students’ self-learning in a
global world with mobility opportunities. It is essential to integrate
existing terminological resources in a unified way (e.g., in UMLS, MeSH,
NCIt, SNOMED CT) and to reduce translation-based misunderstanding
by involving native language speakers directly (as a part of a working
6
International Journal of Medical Informatics 167 (2022) 104860
group) or indirectly (via the social platforms. Therefore, MIMO is a
timely opportunity to develop and improve scientific, technical,
educational, and industrial partnerships in medical informatics and
digital health.
Summary table.
What was already known on the topic:
1. Developing long-term international cooperation is challenging in
conducting research. Creating international partnerships looks more
accessible than ever but can fail due to language differences and
misunderstandings.
2. Managing complex projects in medical informatics and digital health
fields requires a basic but proper understanding of the terminologies
both from the technological and medical sides.
What this study added to our knowledge.
1. MIMO aims to provide a multilingual tool focusing specifically on
health informatics and technologies.
2. The implications for medical informatics and digital health in the
academic and industrial landscapes are essential to develop and to
improve scientific, technical, educational, and industrial partner­
ships in medical informatics and digital health.
3. MIMO can be a pillar component of an environment supporting
health sciences information professionals’ different activities, facil­
itating mutual understanding.
CRediT authorship contribution statement
Arriel Benis: Conceptualization, Methodology, Validation, Formal
analysis, Investigation, Resources, Writing – original draft, Writing –
review & editing, Supervision, Project administration, Funding acqui­
sition. Julien Grosjean: Software, Formal analysis, Investigation, Data
curation, Writing – review & editing, Visualization. Kévin Billey:
Software, Formal analysis, Investigation, Data curation, Writing – re­
view & editing, Visualization. Gustavo Montanha: Validation, Inves­
tigation, Writing – review & editing. Verena Dornauer: Validation,
Investigation, Writing – review & editing. Mihaela Crișan-Vida:
Conceptualization, Validation, Investigation, Writing – review & edit­
ing, Funding acquisition. Werner O Hackl: Validation, Investigation,
Writing – review & editing. Lăcrămioara Stoicu-Tivadar: Conceptu­
alization, Validation, Investigation, Writing – review & editing, Project
administration, Funding acquisition. Stéfan J. Darmoni: Conceptuali­
zation, Methodology, Software, Validation, Formal analysis, Investiga­
tion, Resources, Writing – original draft, Writing – review & editing,
Visualization, Supervision, Project administration.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Acknowledgements
Open Access of this EFMI WG publication has been made possible by
the contribution of EFMI, in accordance with its open access policy.
Funding
This project was partially funded as a part of HosmartAI (Hospital
Smart development based on AI), a European Union’s Horizon 2020
research and innovation program under grant agreement No.
101016834.
A. Benis et al. International Journal of Medical Informatics 167 (2022) 104860

References [27] M.S. Jalali, J.P. Kaiser, Cybersecurity in Hospitals: A Systematic, Organizational
Perspective, J. Med. Internet Res. 20 (5) (2018) e10059, https://doi.org/10.2196/
10059.
[1] L. Stephanie, R.S. Sharma, Digital health eco-systems: An epochal review of
[28] D. Upadhyay, S. Sampalli, SCADA (Supervisory Control and Data Acquisition)
practice-oriented research, Int. J. Inf. Manag. 53 (2020), 102032, https://doi.org/
systems: Vulnerability assessment and security recommendations, Comput. Secur.
10.1016/j.ijinfomgt.2019.10.017.
89 (2020) 101666, https://doi.org/10.1016/j.cose.2019.101666.
[2] N. Rieke, J. Hancox, W. Li, F. Milletarì, H.R. Roth, S. Albarqouni, S. Bakas, M.
[29] D.V. Dimitrov, Medical Internet of Things and Big Data in Healthcare, Healthc.
N. Galtier, B.A. Landman, K. Maier-Hein, S. Ourselin, M. Sheller, R.M. Summers,
Inform. Res. 22 (3) (2016) 156, https://doi.org/10.4258/hir.2016.22.3.156.
A. Trask, D. Xu, M. Baust, M.J. Cardoso, The future of digital health with federated
[30] D.S.W. Ting, L. Carin, V. Dzau, T.Y. Wong, Digital technology and COVID-19, Nat.
learning, Npj Digit. Med. 3 (2020) 119, https://doi.org/10.1038/s41746-020-
Med. 26 (4) (2020) 459–461, https://doi.org/10.1038/s41591-020-0824-5.
00323-1.
[31] CISMeF ([French] acronym for Catalog and Index of French Language Health
[3] N. Kancewicz-Hoffman, J. Pölönen, Does excellence have to be in English?
Resources on the Internet), HeTOP, (2005). https://www.hetop.eu/hetop/en/
Language diversity and internationalisation in SSH research evaluation, in:
(accessed April 29, 2022).
Overview of Peer Review Practices in the SSH, ENRESSH Report, European
[32] J. Grosjean, T. Merabti, B. Dahamna, I. Kergourlay, B. Thirion, L.F. Soualmia, S.
Network for Research Evaluation in the Social Sciences and Humanities. COST
J. Darmoni, Health multi-terminology portal: a semantic added-value for patient
action 15137: pp. 32–41. https://enressh.eu/wp-content/uploads/2017/09/Overv
safety, Stud. Health Technol. Inform. 166 (2011) 129–138, https://doi.org/
iewPeerReviewENRESSH-1.pdf, 2020. (Accessed 3 May 2022).
10.3233/978-1-60750-740-6-129.
[4] S. Schmidt, M. Bullinger, Current issues in cross-cultural quality of life instrument
[33] International Organization for Standardization, ISO 25964–1:2011, International
development, Arch. Phys. Med. Rehabil. 84 (2003) S29–S34, https://doi.org/
Organization for Standardization, 2011. https://www.iso.org/standard/53657.
10.1053/apmr.2003.50244.
html.
[5] M. Kelly (Ed.), Languages After Brexit, Springer International Publishing, Cham,
[34] International Organization for Standardization, ISO 25964–2:2013, International
2018, https://doi.org/10.1007/978-3-319-65169-9.
Organization for Standardization, 2013 https://www.iso.org/standard/53658.
[6] G. Mihalas, J. Zvarova, C. Kulikowski, M. Ball, J. Bemmel, A. Hasman, I. Masic,
html.
D. Whitehouse, B. Barber, History of Medical Informatics in Europe - a Short
[35] S.G.D. Clarke, M.L. Zeng, Standard Spotlight: From ISO 2788 to ISO 25964: The
Review by Different Approach, Acta Inform. Medica. 22 (1) (2014) 6, https://doi.
Evolution of Thesaurus Standards Towards Interoperability and Data Modeling,
org/10.5455/aim.10.5455/aim.2014.22.6-10.
Inf. Stand. Q. 24 (2012) 20–26, https://doi.org/10.3789/isqv24n1.2012.04.
[7] A. Benis, M. Crisan-Vida, L. Stoicu-Tivadar, The EFMI Working Group “Healthcare
[36] J. Grosjean, T. Merabti, N. Griffon, B. Dahamna, S.J. Darmoni, Teaching medicine
Informatics for Interregional Cooperation”: An Evolving Strategy for Building
with a terminology/ontology portal, Stud. Health Technol. Inform. 180 (2012)
Cooperation Bridges, Stud. Health Technol. Inform. 264 (2019) 1907–1908,
949–953, https://doi.org/10.3233/978-1-61499-101-4-949.
https://doi.org/10.3233/SHTI190707.
[37] D.A.B. Lindberg, B.L. Humphreys, A.T. McCray, The Unified Medical Language
[8] N.J. Ogg, M. Sievert, Z.R. Li, J.A. Mitchell, Construction of a medical informatics
System, Methods, Inf. Med. 32 (04) (1993) 281–291, https://doi.org/10.1055/s-
thesaurus, Proc. Symp. Comput. Appl. Med. Care. (1994) 900–904, in: https://
0038-1634945.
www.ncbi.nlm.nih.gov/pmc/articles/PMC2247813/pdf/procascamc00001-0866.
[38] CISMeF ([French] acronym for Catalog and Index of French Language Health
pdf.
Resources on the Internet), CISMeF, (2007). https://www.cismef.org/cismef/pro
[9] M.C. Sievert, D.E. Moxley, N.J. Ogg, T.B. Patrick, in: Thesaurus of Health
jet-cismef/english/ (accessed March 8, 2022).
Informatics, IOS Press, Amsterdam, 1998. https://ebooks.iospress.nl/volume/th
[39] T. Merabti, J. Grosjean, M. Joubert, Aligning Biomedical Terminologies in French:
esaurus-of-health-informatics.
Towards Semantic Interoperability in Medical Applications, in: S. Mordechai (Ed.),
[10] R. Rada, C. Ghaoui, J. Russell, M. Taylor, Approaches to the construction of a
Med. Inform, Lond.), InTech, 2012, https://doi.org/10.5772/37738.
medical informatics glossary and thesaurus, Med. Inform. (Lond.) 18 (1) (1993)
[40] N. Griffon, G. Kerdelhué, L.F. Soualmia, T. Merabti, J. Grosjean, J.-B. Lamy,
69–78, https://doi.org/10.3109/14639239309034470.
A. Venot, C. Duclos, S.J. Darmoni, Evaluating alignment quality between iconic
[11] R. Cornet, L. Stoicu-Tivadar, A. Hörbst, C.L. Parra Calderón, S.K. Andersen,
language and reference terminologies using similarity metrics, BMC Med. Inform.
M. Hercigonja-Szekeres (Eds.), Digital Healthcare Empowering Europeans, Studies
Decis. Mak. 14 (2014) 17, https://doi.org/10.1186/1472-6947-14-17.
in Health Technology and Informatics, 210, IOS Press, Amsterdam, 2015.
[41] Wikipedia contributors, Wikipedia, (2022). https://www.wikipedia.org/ (accessed
[12] A. Benis, M. Crisan-Vida, L. Stoicu-Tivadar, S. Darmoni, A Multilingual Dictionary
March 8, 2022).
for Health Informatics as an International Cooperation Pillar, Stud. Health Technol.
[42] MediaWiki software, API:Main page - MediaWiki, (2022). https://www.mediawiki.
Inform. 262 (2019) 31–34, https://doi.org/10.3233/SHTI190009.
org/wiki/API:Main_page (accessed March 8, 2022).
[13] Europe Direct Information Center, European Day of Languages | Strasbourg
[43] M. Dusenne, K. Billey, F. Desgrippes, A. Benis, S.J. Darmoni, J. Grosjean,
Europe, (2022). https://www.strasbourg-europe.eu/european-day-of-languages/
WikiMeSH: Multi Lingual MeSH Translations via Wikipedia, Stud Health Technol
(accessed March 8, 2022).
Inform 294 (2022) 403–404, https://doi.org/10.3233/SHTI220483.
[14] The World Bank Group, Population, total - European Union | Data, (2022). http
[44] DeCS - Health Sciences Descriptors, (2017). https://decs.bvsalud.org/I/homepage
s://data.worldbank.org/indicator/sp.pop.totl?locations=eu (accessed March 8,
i.htm (accessed March 8, 2022).
2022).
[45] A. Benis, Inivited talk: A multilingual thesaurus for Medical Informatics as an
[15] M. Serva, F. Petroni, Indo-European languages tree by Levenshtein distance, EPL
educational tool, (2020). https://efmi.org/2020/12/04/romanian-society-of-
Europhys. Lett. 81 (6) (2008) 68005, https://doi.org/10.1209/0295-5075/81/
medical-informatics-live-event-14-12-20-adapting-education-during-covid-19-pan
68005.
demic/ (accessed April 30, 2022).
[16] ISO 639-2:1998, (1998). https://www.iso.org/standard/4767.html (accessed
[46] J. Mantas, L. Stoicu-Tivadar, C. Chronaki, A. Hasman, P. Weber, P. Gallos, M.
March 8, 2022).
Crişan-Vida, E. Zoulias, O.-S. Chirila, 31st Medical Informatics in Europe
[17] Google, googletranslate - Docs Editors Help, (2022). https://support.google.
Conference (MIE 2021), held virtually, (2021). https://mie2021.vfairs.
com/docs/answer/3093331?hl=en (accessed March 8, 2022).
com/en/#agenda (accessed April 30, 2022).
[18] Y. Farahsani, I.P. Rini, P.H. Jaya, Google Translate Accuracy in Translating
[47] A. Benis, S.J. Darmoni, L. Stoicu-Tivadar, Medical informatics multilingual
Specialized Language From English to Bahasa Indonesia:: A Case Study on
ontology” at the STC 2021 - Satellite Event “EFMI perspective and contribution for
Mechanical Engineering Terminology, in: Proc. 4th Int. Conf. Sustain. Innov.
the HosmartA H2020 project. https://www.hosmartai.eu/special-topic-co
2020–Social Humanity Educ. ICoSIHESS 2020, Atlantis Press, Yogyakarta,
nference-2021/2838/.
Indonesia, 2021. doi:10.2991/assehr.k.210120.156.
[48] HosmartAI Consortium, HosmartAI – AI for the smart hospital of the future,
[19] B.R. Taira, V. Kreger, A. Orue, L.C. Diamond, A Pragmatic Assessment of Google
(2022). https://www.hosmartai.eu/ (accessed March 8, 2022).
Translate for Emergency Department Instructions, J. Gen. Intern. Med. 36 (11)
[49] Wiktionary, Wiktionary, the free dictionary, (2022). https://en.wiktionary.org/w
(2021) 3361–3365, https://doi.org/10.1007/s11606-021-06666-z.
iki/Wiktionary:Main_Page (accessed March 8, 2022).
[20] O. Bodenreider, The Unified Medical Language System (UMLS): integrating
[50] Official Languages, U. N. (n.d.). https://www.un.org/en/our-work/official-
biomedical terminology, Nucleic Acids Res. 32 (90001) (2004) D267–D270,
languages (accessed April 30, 2022).
https://doi.org/10.1093/nar/gkh061.
[51] A. Benis, O. Tamburis, C. Chronaki, A. Moen, One Digital Health: A Unified
[21] National Library of Medicine, Medical Subject Headings - Home Page, (2022). htt
Framework for Future Health Ecosystems, J. Med. Internet Res. 23 (2) (2021)
ps://www.nlm.nih.gov/mesh/meshhome.html (accessed March 8, 2022).
e22189, https://doi.org/10.2196/22189.
[22] SNOMED International, SNOMED Home page, (2022). https://www.snomed.org/
[52] N. Spicher, R. Barakat, J.u. Wang, M. Haghi, J. Jagieniak, G.S. Öktem, S. Hackel, T.
(accessed March 8, 2022).
M. Deserno, Proposing an International Standard Accident Number for
[23] National Cancer Institute, NCI Thesaurus, (2022). https://ncithesaurus.nci.nih.
Interconnecting Information and Communication Technology Systems of the
gov/ncitbrowser/ (accessed March 8, 2022).
Rescue Chain, Methods, Inf. Med. 60 (S 01) (2021) e20–e31, https://doi.org/
[24] International Council for Harmonisation of Technical Requirements for
10.1055/s-0041-1728676.
Pharmaceuticals for Human Use (ICH), MedDRA, (2022). https://www.meddra.
[53] L.D. Riek, Healthcare robotics, Commun. ACM. 60 (11) (2017) 68–78, https://doi.
org/ (accessed March 8, 2022).
org/10.1145/3127874.
[25] World Health Organization, International Classification of Diseases (ICD), (2022).
[54] C. Wu, P. Wu, J. Wang, R. Jiang, M. Chen, X. Wang, Ontological knowledge base
https://www.who.int/standards/classifications/classification-of-diseases
for concrete bridge rehabilitation project management, Autom. Constr. 121 (2021)
(accessed March 8, 2022).
103428, https://doi.org/10.1016/j.autcon.2020.103428.
[26] V.N. Vahia, Diagnostic and statistical manual of mental disorders 5: A quick glance,
[55] M. Monteiro, E. Keating, Managing Misunderstandings: The Role of Language in
Indian J. Psychiatry. 55 (3) (2013) 220, https://doi.org/10.4103/0019-
Interdisciplinary Scientific Collaboration, Sci. Commun. 31 (1) (2009) 6–28,
5545.117131.
https://doi.org/10.1177/1075547008330922.

7
A. Benis et al.
[56] S. von Danwitz, Managing inter-firm projects: A systematic review and directions
for future research, Int. J. Proj. Manag. 36 (3) (2018) 525–541, https://doi.org/
10.1016/j.ijproman.2017.11.004.
[57] J. Watermeyer, Z. Thwala, J. Beukes, Medical Terminology in Intercultural Health
Interactions, Health Commun. 36 (9) (2021) 1115–1124, https://doi.org/10.1080/
10410236.2020.1735700.
International Journal of Medical Informatics 167 (2022) 104860
[58] P.W. Teunissen, A. Atherley, J.J. Cleland, E. Holmboe, W.C.Y. Hu, S.J. Durning,
H. Nishigori, D.D. Samarasekera, L. Schuwirth, S. van Schalkwyk, L.A. Maggio,
Advancing the science of health professions education through a shared
understanding of terminology: a content analysis of terms for “faculty”, Perspect,
Med. Educ. 11 (2022) 22–27, https://doi.org/10.1007/s40037-021-00683-8.