On the Use of an Ontology to Improve

the Interoperability and Accesibility

of the Electronical Health Records (EHR)

Belen Prados Suarez

, Carlos Molina

, Miguel Prados

and Carmen Peña

Software Engineering Department, University of Granada, Granada, Spain

Computer Science Department, University of Jaén, Jaén, Spain

University Hospital “San Cecilio”, Computing Department, Granada, Spain

Abstract. In this paper we present a proposal to conceptualize the EHR, based

on the semantic description of the information, according to the documentary

structures and the clinical aspects of the EHR contents. Our aim here is to per-

form a formalization with a double purpose: on one hand to enable the interope-

rability; on the other hand, to improve the accessibility to the EHR, according

to clinical or assistance contexts, provi-ding the clinical data retrieval system

with flexibility and operativity. To this purpose we propose the use of an On-

tology to represent this conceptualization, and include properties and relations

between the components of the EHR.

1 Introduction

Every day more the Electronic Health Record (EHR) is an extended reality in the

majority of the Hospitals, with different degrees of development. It has opened the

access to new uses of the EHR, optimized and with more benefits for the medical

acti-vity. However, new problems and perspectives have also arisen, related to the

management of the clinical information [1].

As the use of the EHR spreads over the different medical specialities and assis-

tance acts, it must integrate more documents and information items, from different

sources and types. It is unavoidable to think on the risk that the EHR runs of becom-

ing as unmanageable as the old health records in paper: with such a quantity of in-

formation and documents, the access to concrete data items required in relatively

simple situations can be really difficult.

Another main problem is the interoperability, with the aim is to communicate and

make possible the understanding between different models of EHR from different

hospitals and providers. The ISO 13606 [2] regulation establishes the basis and gen-

eral framework of the semantic interoperability model [3], to allow the univocal in-

terpretation of the information transmitted during the capture of the context where it

was generated. The ISO 13606 regulation proposes a dual model where the first

model is the reference model and the second one is the archetypes model. Both of

them will be commented later.

In addition to the above mentioned problems, there are also several important is

Prados Suárez B., Molina C., Prados M. and Peña C..

On the Use of an Ontology to Improve the Interoperability and Accesibility of the Electronical Health Records (EHR) .

DOI: 10.5220/0003352500730081

In Proceedings of the International Workshop on Semantic Interoperability (IWSI-2011), pages 73-81

ISBN: 978-989-8425-43-0

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

sues that must also be addressed and solved, like:

• The Mobility: The use of mobile devices (tablet PC, PDAs, …) requires agile and

summarized navigation models on the EHR.

• The Contextualization: The contextual use of the information would provide the

doctor with information pertinent to the assistance act where he/she is involved.

• Access Focusing: The idea is to allow the navigation through the EHR directing

the search according to a semantic purpose.

• The personal access of the owner: Every day more the patients demand the access

to their clinical data since, as owner of them, they have the right to access them.

However, at this moment nor the systems are ready for this purpose nor the citizen

have the technical knowledge to use them.

The bibliography reviewed shows a great concern about these problems of the EHR

systems, specially the interoperability one. In particular about the use of Ontologies

as a means to represent models capable of understand and communicate to each other

[10]. However, though in the literature most of the proposals are mainly focused on

the interoperability and its tools, they do not propose explicitly applications towards

the accessibility, use and management of the EHR at a local level.

In this paper we offer a different point of view, more focused in this latter line,

offering different alternatives of use of the EHR for the doctors, making more

efficient the accessibility to the information needed. It is quite important since the

great volume of documents and information contained in the EHRs, is so extensive

that usually one of the main causes of complains from the users is related to the

difficulties to navigate through them.

It all have lead us to think that the EHR can be considered as a universe of know-

ledge that can be conceptualized in such a way that each individual information item

can be defined in a semantic family, according to its properties and relations with

other information items [6].

In addition, this paper provides with a novel point of view, applying the Ontology

based representation models to make easier the use and navigation.

It is specially interesting, since opens the possibility of performing semantic re-

trieval of information, through the construction of Agents that interpretate the queries

and allow the access to the pertinent data items.

Our proposal then, is an approach to the conceptualization of the universe of

clinical data contents of the EHR, through the use of an Ontology, based on our own

EHR model. Starting from this conceptual formalization and on the basis of the gen-

erated knowledge we can construct “agent procedures”. These procedures approach

the user to the information items that, due to their semantic value, can be more useful

according to the access model.

2 Background

Before presenting our proposal we are going to point some notes about the frame-

work from which we have started our research, specially the information system and

the legal framework.

2.1 System Used

To perform our proposal we have based on the system “ARCHINET”, the EHR sys-

tem developed and implemented in the Universitary Hospital San Cecilio of Granada

(Spain) [1]. This system was implemented 10 years ago, is continuously improved

and counts on around 1.000.000 of EHRs. It is organized according to the highest

level structures: “assistance episodes” and “pathological processes”.

The EHR (Figure 1) is a set of documents (e.g. cardiograms, analytical tests, etc.),

and each of them may consist on a set of data groups (e.g. some variable in an ana-

lytical test, each of the images in a TAC, etc).

Health Record

Asistencial episode 1

Document 1

Data Group 1

Data Group 2

DATA

Document N

Data Group 1

DATA DATA

DATA

Document 2

Data Group 1

DATA

Fig. 1. Example of EHR structure.

In addition we have also made use of the “access base” that registers all the ac-

cesses to the system, information items acceded or modified and the assistance con-

text where the user was involved.

To define the Ontology’s structures, the Protégé tool has been used, allowing also

the operations of validation, representation and translation to descriptive languages.

2.2 Legal Framework

As mentioned above, the ISO 13606 regulation proposes a dual model where the first

model is the reference model and the second one is the archetypes model.

The reference model is used to represent the structure of the clinical data of a con-

crete model like, as an example, the model of a given hospital. It is based on a class

called “structure” that gives rise to the following hierarchy of members:

• Folder: It represents the divisions at the highest level inside the extracts of the

clinical history.

• Composition: It is the set of annotations related to a unique given clinical session

or document.

• Sections: They are groupings in a clinical session.

• Entry: Each one represents a clinical observation or a set of them.

• Cluster: It is used when the representation of a unique observation or action re-

quires a complex data structure, like a list, a table or a temporal series.

• Element: It contains a unique value that must be instance of some of the types

defined by it.

The second model sets the Archetypes [4], [5] as a way to define the clinical concepts

managed by the system. The archetypes are definitions of sets of clinical information

items, that have a concrete clinical meaning; and they are created using the compo-

nents defined in the ISO 13606.

However this regulation just sets the basis and general description on which eve-

rything is opened and must be concreted, which is what we do in this paper.

3 Ontology Proposed

To materialize the general models described in the previous section, we propose and

describe in this section an ontology.

3.1 General Description

The EHR structure’s design itself implies the existence of a categorization according

to the semantic classes of the documentary organization, and also to the assistance

part. As an example, the documents are classified by their types, and the data are

organized regarding their clinical orientation inside the document. In addition these

data items are organized according to assistance acts and medical specialities. Based

on it we make the formalization through the ontology.

To choose the components of the Ontology we have used two criteria. On one

hand, the documentary criteria, that gives rise to the classes that structure and define

the set of documents included in the EHR. On the other hand, we approach a clinical

criteria determinating the categories related to the clinical processes and concrete

pathologies, and even the assistance context on which the information is used.

Nevertheless, the semantic universe must be manageable and easily formalizable,

so we have avoided to define categories that are not clearly useful to reach our final

targets, previously indicated.

3.2 Semantic Categories in the Ontology

We have analyzed the semantic categories to define the Ontology. The main purpose

was double: on one hand, to categorize every kind of data that could be found in the

EHR, on the other hand to respect the information structures defined in our EHR

model. Whit it the classes defined in the Ontology are:

Structure Model (EHR-EXTRACT): This class corresponds to the structure defi-

ned in the ISO 13606, and has the members Folder, Composition, Section, Entry,

Cluster and Element.

Document: A document can be considered as any grouping of data with a common

purpose, nested regarding a clinical action or observation. The documents are hierar-

chized depending on whether they are “general”, “of process”, “of medical special-

ity”, “of nursery”, “surgical” or “logistical”. Hence, this class can be considered as

the fundamental logical grouping of the organization of the information in the EHR.

With this class the EHR can be organized according to assistance acts (admissions,

consultations, emergencies,…) or to pathological processes, always grouping docu-

ments. Each document may contain different sections of contents, and each section

has its own entries, clusters and elements as concrete data in the document.

Assistance Process: These processes define the clinical pathology environments,

previously set, on which sequences of clinical actions are pre-established. As an ex-

ample, we have the “diabetes process”, “cataract process”,… Here we have focused

on the pathologies with well defined processes, since not all the pathologies have

them. The members of this class represent the different pre-established actions for

each process.

Data Type: They can be considered as texts, encoded data, magnitudes that include

rations, intervals, lengths, durations, graphs, images, signals, dates and so on.

Observation Type: The aim of this class is to qualify the data item according to its

source: if it is a subjective observation, an objective result of an analysis, a protoco-

lyzed observation, a related fact or a chronological action, among others.

Assistance Procedure: It contains the references to the diagnosis methods, explora-

tions, sources of knowledge, technological support, and any other source of data. As

an example, we have electromedical explorations (electrocardiogram, electroencepha-

logram,…), radiological explorations (RMN, TAC, conventional radiology,…), and

direct observation, among others.

Clinical Context: It is related to the variety of situations or states of an assistance act,

like a revision consultation, a postsurgical consultation, an admission, an emergency

assistance or a ward checkup. These contexts are obviously classified according to the

medical speciality and, in some cases, to sub-speciality and process.

Assistance act: It determines the origin of the assistance procedure (admission, con-

sultation, emergency,…).

Agent: This class is used to define the kind of professional that is involved in the act,

locating him/her in the corresponding service and professional category (doctor,

nurse, assistant,…).

Archetype: We use the internally defined archetypes and those other defined by the

different research groups working on the interoperability of the EHR [7].

Application: This class captures the variety of functional applications from different

providers and the specific tools, that the clinical workstations entail and must be inte-

grated. These applications are complimented in the system ARCHINET by means of

its own and specific functionalities, with a logistical or departmental character. Some

examples are the application of medicine and unidosis management, the application

of analytical requests management, or the emergency monitoring. Some of these ap-

plications may derive clinical data towards the EHR.

ICD-10 Hierarchy: It reproduces the class structure in this international classifica-

tion [8]. We have chosen this classification since it is the most habitually used in the

Hospital for diagnosis encoding.

Data Model: It is a class of internal use for the procedures of computing agents. Its

aim is to reproduce the data model starting from its logical modelling and down to

reach the physical Data Base model of the EHR. This is how the tables stored in the

data base are described. The instances of this class are each of the individual data

(columns). The hierarchy of this class shows the typology of these structures: move-

ment tables, primary tables, history tables, etc.

3.3 Properties in the Ontology

Regarding the properties in the Ontology, their purpose is to create sets of restrictions

based on the taxonomical relation between classes, in such a way that each possible

entry in a EHR has a semantic map to contextualize its use, and hence its relations to

other elements in the EHR.

This way, as an example, the entry “anaesthesia type” belongs to the document

“anaesthesia sheet”, is a data type of restricted values, and is part of the context “in-

trasurgical information” and of the assistance procedure “Anaesthesia”. In addition,

it is characteristic of the assistance acts “admission”, “emergency” and “surgical day

hospital”. Its agent profile is “anaesthetist doctor” and it is considered as related to

the archetype “anaesthetic report”.

Fig. 2. OWL Based Relationships Between Classes and their Properties.

To allow the creation of these “semantic contextual maps” in the Ontology we ha-

Type of information

. . . .

Cluster

Controlled values

Document

. . . .

Elect. Document Pre-anaesthetic

Section Ventilation

Ventilation

Agent Æ . . . . Æ Anesthetist doctor

Process

. . . .

Surgical

General procedure

Procedure Æ . . . Æ Electro medical observation

Application

. . . .

Carmenes

Archinet

Specialist

. . . .

Anesthesiology

Archetype

. . . .Complementary study

Pre-anaesthetic

study

ve defined complex relations between classes and their corresponding attributes and

restrictions.

The process to create these relations is quite complex. However, it is easier using

the information stored in the EHR Base and in the Access Base, and also referencing

the data model itself. Doing it most of them can even be automatically generated.

In the Ontology we have also included implicit properties for concrete classes like

the “character of a document” (confidential or open), or the “type of document”

(gene-ral, of speciality, of process, logistical,…).

Finally, we must remark that the definition of the Ontology is not a closed topic,

but a continuous process that, depending on the experiments, we widen or modify.

4 Results

The creation of the Ontology provides a Knowledge Base formalized with structures

that the computing procedures can use to answer the query processes performed on

the EHRs [9], and opens the possibility of using new accessibility models to the EHR.

Concretely, it makes possible the conceptual accessibility to the data in the EHR,

what opens the path towards the interoperability between EHR systems, since it pro-

vides the system with the capability to semantically interpretate the clinical data re-

trieval pro-cessess. In addition, it sets the basis for the next uses of the information

and the system:

Contextual Use: to allow the doctors to have the information really needed for the

assistance activity in which he/she is involved, acceding just to a determined context.

This way superfluous or not pertinent information items are avoided, as well as com-

plex accesses with the navigation systems.

Restricted Navigation: used in the cases where only some concrete information

items are needed, avoiding the unuseful navigation through acts, processes and

documents with no interest to the search purposes.

Limited Navigation for Mobile Devices: The navigation through the contents of the

EHR is quite difficult in mobile devices, since their screens set a very limited repre-

sentation capability, especially for complex menus. In this case, the information pre-

sented can be initially focused according to a given work environment, like the medi-

cal speciality, the assistance act to be performed, the process or the assistance proce-

dure. All of them set an environment to which the system can give a response depend-

ing on the information relevant to it.

Ontology Navigability: Traditionally there have been discrepancies regarding the

different ways to show the documental organization of the EHR. Some times it is

necessary to organize them according to assistance acts, whereas in other cases the

organization according to processes is preferred. In our case the user can choose, with

the scheme of classes that the Ontology provides, allowing him/her to design of

his/her own navigation model.

Interoperability: It is easier to reach with the Knowledge Base provided, making

possible the understanding with other formalized models, especially with the Refer-

ence and Archetypes models defined in the ISO 13606.

Access According to the Semantic Valuation: It makes possible the direct access to

elements contained in the EHR, using the terminology in the Ontology.

As a summary, the Ontology conceptualizes our model of EHR, opening the access to

a great variety of opportunities to develop computing procedures to make easier the

use, control and availability if the EHR.

To our best knowledge, there are some proposals of ontologies for contextualized

access in others fields (e.g. e-Goverment, business context [13]) but none for EHR

access so a comparison with our proposal is not possible.

5 Conclusions

In this paper we have made several proposals:

1. We have presented a semantic conceptualization model for an EHR system, that

offers a number of utilities towards three purposes: the interoperatibility, the accessi-

bility and the mobility.

2. We have proposed concrete accessibility models for the EHR, as a practical appli-

cation of the design proposed.

3. The proposed design can be generally and widely applied, independently of the

documentary structure, the technological support or the development degree.

4. The Ontology provides a formalized Knowledge Base that allows the

development of computing procedures with several purposes, from analytical to the

accessibility, opening the path to new alternatives to the traditional navigation and

access procedures to the EHRs.

However, we assume that this work is just the “starting point” for future develop-

ments and for the creation of computing procedures, more or less “intelligent”, to be

used as user interfaces for the EHR.

At this moment the work carried out has only been limited to the design and con-

struction of the Ontology, and must be continued with the production of the corre-

sponding computing agents.

Regarding this research line it is not finished, since it is just in an experimental

phase, and is opened to modifications in the Ontology design, depending of the re-

sults obtained. In the present phase, we are working on the development of computing

interfaces procedures and in the automation of the generation of classes and proper-

ties of the Ontology from the information stored in the system.

Finally, we must indicate that the experimental results obtained up to this moment

lead us to consider quite viable the implementation at a general level.

References

1. Prados M. and Peña M. C., Sistemas de Información hospitalarios. Organización y gestión

de Proyectos. EASP (Escuela Andaluza de Salud Pública), Granada, (2003).

2. UNE-EN 13606-1:2007. Informática sanitaria. Comunicación de la historia clínica electró-

nica. Parte 1: Modelo de referencia. UNE-EN 13606-2:2007. Informática sanitaria. Comu-

nicación de la historia clínica electrónica. Parte 2: Arquetipos. (2007)

3. Maldonado Segura J. A., Moner Cano, D., et al. Semantic Upgrade and Normalization of

Existing EHR Extracts. Proceedings of the 30th Annual International Conference EMBC,

pp. 1466-1469. ISBN: 978-1-4244-1815-2. ISSN: 1557-170X.

4. P. Serrano, D. Moner, T. Sebastian, J. A. Maldonado, R. Navalón, M. Robles

, Á. Gómez.

Representación De Estructuras De Datos Clínicos Mediante Arquetipos Y Terminologías.

Inforsald (2009).

5. V. Bicer, O. Kilic, A. Dogac, and G. B. Laleci. Archetype-Based Semantic Interoperability

of Web Service Messages in the Health Care Domain. Journal on Semantic Web & Infor-

mation Systems, 1 (4), 1-23. (2005).

6. C. Golbreich, O. Dameron, B. Gibaud, and A. Burgun. Web ontology language require-

ments w.r.t expressiveness of taxonomy and axioms in medicine. In 2nd International se-

mantic web conference, ISWC 2003, Sanibel Island, Florida, US, October 20, 2003 pro-

ceedings. Berlin: Springer. (2003).

7. Sundvall E, Qamar R, Nyström M, Mattias F, Hakan P, Ahlfeldt H, Rector AL. Integration

of Tools for Binding Archetypes to SNOMED CT. BMC Medical Informatics and Decision

Making. 8, S7 (2008).

8. World Health Organization; International Classification of Diseases (ICD):

http://www.who.int/classifications/icd/en/

9. J. Mei y E. P. Bontas. Reasoning Paradigms for SWRL-Enabled Ontologies. In Protégé

With Rules Workshop: 8th International Protégé Conference, Madrid, Spain, (2005).

10. Garde S, Knaup P, Hovenga E, Heard S. Towards semantic interoperability for electronic

health records. Methods Inf.Med. 46, 332-343 (2007).

11. JA. Maldonado, D. Moner, D. Boscá, C. Angulo, I. Abad, D. Pérez, P. Serrano, E. Reig, M.

Robles. LinkEHR-Ed: Una herramienta para la estandarización de la historia clínica

electrónica. Libro de comunicaciones del Inforsalud. Pag. 317-322 (2008).

12. M. O’Connor, H. Knublauch, S. Tu, B. Grosof, M. Dean, W. Grosso, y M. Musen. Sup-

porting Rule System Interoperability on the Semantic Web with SWRL. In Proceedings of

the 4th International

13. Hamdi Chaker, Max Chevalier, Chantal Soule-Dupuy, and Andre Tricot. Improving infor-

mation retrieval y modelling business context. In Advances in Human-Oriented and Perso-

nalized Mechanisms, Technologies and Services (CENTRIC), 2010 Third International

Conference on, pages 117 –122, 2010.