An Automatic System for Helping Health Consumers

to Understand Medical Texts

Marco Alfano

1,2

, Biagio Lenzitti

, Giosuè Lo Bosco

1,3

and Valerio Perticone

Department of Mathematics and Computer Science, University of Palermo, via Archirafi n.34, Palermo, Italy

Anghelos Centre on Communication Studies, via Pirandello n.40, Palermo, Italy

Euro-Mediterranean Institute of Science and Technology, via E. Amari n.123, Palermo, Italy

Keywords: E-Health, Public Health, Healthcare Management Systems, Patient Empowerment, Plain Language,

Consumer Health Vocabulary, Infobutton, Electronic Health Record, Personal Health Record.

Abstract: Medical texts (reports, articles, etc.) are usually written by professionals (physicians, medical researchers,

etc.) who use their own language and communication style. On the other hand, these texts are often read by

health consumers (as in the case of medical reports) who do not have the same skills and vocabularies of the

experts and can have difficulties in text comprehension. To help a health consumer in understanding a

medical text, it would be desirable to have an automatic system that, given a text written with medical

(technical) terms, translates them in simple or plain language and provides additional information with the

same kind of language. We have designed such a system. It processes online medical documents and

provides health consumers with the needed information for their understanding. To this end, we use a

medical vocabulary for finding the technical terms in the medical texts, a consumer health vocabulary

(CHV) for translating the technical terms into their consumer equivalents and a health-consumer dictionary

for finding supplementary information on the terms. We have built a prototype that processes Italian

medical reports and uses infobuttons next to the technical terms for allowing easy retrieval of the desired

information.

1 INTRODUCTION

Medical texts (reports, articles, etc.) are usually

written by professionals (physicians, medical

researchers, etc.) who use their own language and

communication style. On the other hand, these texts

are often read by health consumers (as in the case of

medical reports) who do not have the same skills and

vocabularies of the experts and can have difficulties

in text comprehension (Keselman and Slaughter,

2007; Seedorff and Peterson, 2013).

A language that can be understood by almost

anyone, made of common every-day terms, should

be possibly used in writing medical content that also

goes to non-experts. As an example, plain language

is used by the US Government for improving its

communication with the public (Plain

Language.com, 2014). It is defined as a kind of

language that audience can understand the first time

they read it. Simple English Wikipedia is another

example of content developed using 2,000 common

English words (Simple English Wikipedia, 2014).

Of course, it is not always easy for the medical

experts to write in plain language because of the risk

of loosing precision and the time often required to

simplify concepts.

To help a health consumer in understanding a

medical text, it would then be desirable to have an

automatic system that, given a text written with

medical (technical) terms, translates them in plain

language and provides additional information with

the same kind of language. In this way, the user is

not required to increase his/her knowledge to a level

where he/she is able to understand the text being

examined but knowledge is somehow brought back

to the user level so greatly facilitating his/her

comprehension. Moreover, it can help the

communication between experts and non-experts

(e.g., physicians and patients or scientists and

laymen) by providing a sort of two-way translation

of terms.

In this work, we describe a system that, given an

online medical text, finds the technical terms,

translates them in a plain language and provides

622

Alfano M., Lenzitti B., Lo Bosco G. and Perticone V..

An Automatic System for Helping Health Consumers to Understand Medical Texts .

DOI: 10.5220/0005283606220627

In Proceedings of the International Conference on Health Informatics (HEALTHINF-2015), pages 622-627

ISBN: 978-989-758-068-0

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

additional information for them (in plain language)

greatly facilitating text understanding. To this end,

we use a medical vocabulary for finding the

technical terms in the medical text, a consumer

health vocabulary (CHV) for translating the

technical terms into their equivalents for consumers

and a consumer dictionary for finding

supplementary information on the terms. We have

developed a prototype that processes Italian medical

reports and uses infobuttons next to the technical

terms for allowing easy retrieval of the desired

information.

Our system, to the best of our knowledge,

constitutes one of the first solutions in this field. In

fact, infobuttons have been used for years mainly to

support clinicians’ decisions and only recently they

have been used to bring information to patients

(Baorto and Cimino, 2000; Kemper, 2010).

Moreover, a couple of systems that present some

similarities with our system are described in (Zeng-

Treitler and Goryachev, 2007) and (Kandula, 2010).

However, they replace the medical terms with the

consumer ones and add further information within

the text, altering the original text. Our system

provides the consumer information ‘aside’ the

original text leaving it intact. Moreover, our system

provides term definitions in consumer language

whereas the other two systems do not.

The paper is organized as follows. The second

section describes the resources that help health

consumers to understand a medical text. The third

section describes the infobuttons that can be used for

automatic retrieval of health information. The fourth

section describes the architecture and

implementation details of the system we have

developed together with some practical use. The

final section presents some conclusions and future

work.

2 MEDICAL VOCABULARIES,

THESAURI AND

DICTIONARIES

As said in the Introduction, when a health consumer

reads a medical text written by a professional he/she

has often difficulties in understanding it because of

the used terms and language. He/she may then need

some external help to understand the technical

terms, find familiar synonyms and get additional

information. This external help comes in the form of

different resources (online or not) such as

vocabularies, dictionaries and thesauri:

 a ‘medical vocabulary’ is a selective list of

words and phrases used in the medical field;

it can be used to find the technical terms in a

medical text;

 a ‘medical-consumer thesaurus’ contains

synonyms and antonyms of medical terms; it

can be used to find plain synonyms of the

technical terms;

 a ‘health-consumer dictionary’ gives

information about the meaning of the words;

it can be used to find additional information

on the technical terms.

In some cases, a single resource can have

multiple functionalities, e.g., it can contain both

definitions and synonyms. Of course, there are

numerous resources of each type. In the next

subsections we briefly describe the resources we

have used for building the system presented in

Section 4 that deals with Italian reports even though

the proposed methodology can be applied to any

language.

2.1 Medical Vocabularies

Medical vocabularies, as said above, are selective

lists of words and phrases used in the medical field.

Usually they are created for the professionals and,

for this reason, they contain all the technical terms

used in the medical field. Usually, health consumers

use these resources when they do not find the

medical terms in the consumer dictionaries or want

more technical information on the searched term.

The ‘Unified Medical Language System

(UMLS)’ is a large collection of multilingual

vocabularies that contains information about

biomedical and health related concepts created and

maintained by the US National Library of Medicine

(Unified Medical Language System, 2014). It

mainly uses a ‘Concept Unique Identifier – CUI’ (a

unique identifier for each concept) to create a

mapping among these vocabularies and thus allows

translation among the various terminology systems.

It may also be viewed as a comprehensive thesaurus

and ontology of health and biomedical concepts.

The ‘Dizionario di Medicina e Biologia’ by

Zanichelli is a bilingual (italian and english)

vocabulary and dictionary written by medical

experts and used not only by technical translators,

but also by professionals, physicians and health

executives (Dizionario di Medicina e Biologia,

2014). Each entry contains an encyclopaedic section

with English translation and Italian explanation and

it is frequently updated.

AnAutomaticSystemforHelpingHealthConsumerstoUnderstandMedicalTexts

623

2.2 Medical-Consumer Thesauri

Consumer terms are not usually well covered by the

existing medical vocabularies, which mostly

represent the language of health professionals (Zeng

and Tse, 2006). Indeed, expressions used by

consumers to describe health-related concepts and

relationships among such concepts frequently differ

on multiple levels (i.e., syntactic, conceptual and

explanatory) from those of professionals. As a

consequence, consumer health vocabularies (CHVs)

have been created for translating medical terms and

concepts in their equivalent for consumers and they

can be very useful for translating medical (technical)

terms in consumer ones (Zielstorff, 2003).

One of the best known examples of CHV is the

‘Open Access Collaboratory Consumer Health

Vocabulary (OAC-CHV)’ created and maintained by

the Consumer Health Vocabulary Initiative

(Consumer Health Vocabulary Initiative, 2014). It is

a relationship file that links commonly used terms to

associated medical terminology represented by the

UMLS. The OAC-CHV focuses on expressions and

concepts that are employed in health-related

communications from or to consumers.

The OAC-CHV contains around one hundred

and sixty thousand rows (one for each term) and

different fields among which:

- ‘Term’: The term as found in the text;

- ‘Concept Unique Identifier’ (CUI): The

unique identifier of a concept as found in the

UMLS;

- ‘CHV Preferred Name’: The preferred

consumer term as defined in the Consumer

Health Vocabulary;

In Section 4, we will show how we have used

these fields for translating technical terms in

consumer ones.

2.3 Health-consumer Dictionaries

There are many sites created for health consumers

that contain health and medical information suitable

to general users. These sites usually contain health

and medical dictionaries specifically created for

health consumers and then use a language that can

easily be understood by them. Examples of online

dictionaries in English are WebMD (WebMD, 2014)

and MedlinePlus (MedlinePlus, 2014)] and online

dictionaries in Italian are Ok Salute (OK Salute,

2014) and Dizionario della Salute (Dizionario della

Salute, 2014).

3 INFOBUTTONS FOR

AUTOMATIC RETRIEVAL OF

INFORMATION

Infobuttons are context-sensitive links usually

inserted in online medical texts, as the ones found in

a Clinical Information System (CIS), Electronic

Health Record (EHR) or Personal Health Record

(PHR), and allow easy retrieval of relevant

information (Cimino, 2007). The online medical text

shows a button next to some of its parts such as

diagnoses or prescriptions. When clicked, the

infobutton creates a query based on the context of

the interaction and sends it to electronic health

information resources retrieving information that

helps understanding or completing the text being

read (Infobuttons, 2014).

The Health Level Seven (HL7) Organization has

created a standard for infobuttons, the “Context

Aware Knowledge Retrieval Application

(Infobutton), Knowledge Request”, that provides a

standard mechanism for clinical information systems

to submit knowledge requests to knowledge

resources. The specification also defines a shared

context information model to be implemented by

EHR/PHR systems and knowledge resources

(Infobuttons, 2014; HL7 Product Infobuttons, 2014).

Based on the context, which includes characteristics

of the patient, provider, care setting, and clinical

task, infobuttons anticipate clinicians’ and patients’

questions and provide automated links to resources

that may answer those questions. For example, an

infobutton displayed in the context of a patient’s

problem list may allow a clinician to retrieve

treatment guidelines on a specific condition as well

as relevant patient education material to share with

the patient (HL7 Version 3 Implementation Guide,

2014).

As said above, infobuttons have been used for

years mainly to support clinicians’ decisions

regarding diagnosis and orders for tests and

medications. More recently, infobuttons have also

been used to trigger relevant and helpful information

to patients (Baorto and Cimino, 2000; Kemper,

2010). We use infobuttons in a similar way, as we

show in Section 4, i.e., to bring useful information to

health consumers by automatically providing

consumer synonyms and term definitions.

HEALTHINF2015-InternationalConferenceonHealthInformatics

624

4 AN AUTOMATIC SYSTEM FOR

HELPING COMPREHENSION

OF MEDICAL TEXTS

As seen in Section 2, a health consumer reading a

medical text will often use some additional resources

to find the technical terms, their synonyms and

related information so, ultimately, to understand the

whole text. However, this approach assumes that the

additional resources are readily available and this is

not always the case. Moreover, the resources are not

accessed in an established order leading to a

disorganized process that can be time consuming

and bring an information overload. It is then

important to develop a process that uses the

additional resources in a coherent and efficient way,

even better if this process is automatized. The final

objective of this process is the translation of the

technical terms of a medical text to plain language

and the provision of additional information for

improving its understanding.

The basic steps of this process can be

summarized as follows:

1. Take a medical text (possibly online)

and find and highlight the technical

terms (words or combinations of words)

by using the medical vocabulary.

2. Translate the highlighted technical terms

to non-technical, or plain, terms with the

medical-consumer thesaurus.

3. Finally, provide additional plain

information with the health-consumer

dictionary.

By doing so, the user will have everything at hand

by dealing with a single document that contains all

the useful information for its understanding.

4.1 System Architecture and

Implementation

In the frame of a collaboration with some Italian

hospitals for providing advanced tools to health

consumers, we have developed a system that

automatically finds the technical terms in an online

medical document, translates them in plain or

consumer terms and provides additional information

in plain language. The architecture of the system is

shown in Figure 1.

The HIGHLIGHT module takes as input an

arbitrary text and, using a medical vocabulary,

highlights all the technical terms. The MAP module

connects the technical terms previously found to

their equivalent consumer terms by using a thesaurus

and puts an infobutton with a question-mark icon (?)

next to the item for which a consumer translation

exists. When clicked, the infobutton will show the

consumer translation of the term in a tooltip near the

item. The DEFINE module provides a description of

the term retrieved by a consumer dictionary. It will

put an infobutton with an information icon (i) next to

each item for which a consumer definition exists

and, when clicked, it will show the definition in a

separate frame under the main text. This definition

will also be processed by the whole system and

transformed in an annotated hypertext that highlights

the technical terms and add the related infobuttons

so to allow the user a deeper analysis and

navigation.

Figure 1: System architecture.

Since our system has to work with medical texts

written in Italian, the HIGHLIGHT module uses, as

medical vocabularies, the Italian vocabularies

present in the UMLS, namely the Italian versions of

the ‘Medical Subject Headings’ (MeSH), the

‘International Classification of Primary Care’

(ICPC), the ‘Medical Dictionary for Regulatory

Activities Terminology’ (MedDRA) and the

‘Metathesaurus Version of Minimal Standard

Terminology Digestive Endoscopy’ (MTHSMS) for

a total of around 150,000 entries. Moreover, it also

uses the ‘Dizionario di Medicina e Biologia’ by

Zanichelli which has around 60,000 entries.

The MAP module uses, as a thesaurus, the Open

Access Collaboratory Consumer Health Vocabulary

(OAC-CHV) seen in Section 2.2. The mapping from

technical to consumer terms is accomplished by

means of the Concept Unique Identifier (CUI) when

available (Keselman, 2008). This is the case of the

technical terms that are found in the UMLS.

Notice that, since there is no Italian CHV

available, beside the ‘Italian Consumer-oriented

Medical Vocabulary’ (ICMV) that only contains a

few items (Italian Consumer-oriented Medical

Vocabulary, 2014), we proceeded in translating the

OAC-CHV (with its 160,000 entries) from English

to Italian. In particular we have translated the

‘Terms’ and the ‘CHV Preferred Names’ elements

described in Section 2.2. This has been done by

using the UMLS multilingual vocabularies, the

AnAutomaticSystemforHelpingHealthConsumerstoUnderstandMedicalTexts

625

English-Italian translation of the ‘Dizionario di

Medicina e Biologia’ and Google Translator. To

improve the effectiveness of the Google translation,

we have replaced a term translated by Google with a

UMLS term with the same CUI when their

‘distance’, computed with the Levenshtein distance

algorithm (Levenshtein, 2014), was short.

The DEFINE module uses the Italian consumer

health dictionaries ‘Ok Salute’ (around 8,000

entries) and ‘Dizionario della Salute’ (around 6,000

entries) to find the definitions of the technical terms.

When a term is not found in these dictionaries, the

definition is searched in the ‘Dizionario di Medicina

e Biologia’ (around 60,000 entries) although its

definitions have a more technical nature. If a

definition is not found in any of the dictionaries, we

use the ‘Google define’ keyword to provide a

definition in any case.

4.2 Practical Use

We have implemented the system and created a

prototype that has been tested with some Italian

medical reports. It can be found at the address

http://math.unipa.it/simplehealth/simple.

Figure 2 shows a snapshot of the prototype that

presents an input form where it is possible to insert

any (Italian) medical text and process it or choose

among some Italian medical reports. In the case of

Figure 2, we have loaded a magnetic resonance

(MRI) report.

Figure 2: Input form for medical text processing.

Figure 3 shows the same medical report where the

technical terms are highlighted after being processed

through our system. Next to each technical term, we

find one or two infobuttons, i.e., a question mark-

icon for the consumer translation of the term (in a

tooltip next to the term) and an information-icon for

its definition (in a frame below). In particular, The

word encefalo (encephalon in English) is selected

and its consumer translation, cervello (brain in

English), is shown together with its explanation in

the below frame.

Figure 3. Medical report with highlighted technical terms.

From a first analysis, our system is capable of

finding (and highlighting) most of the technical

terms present in the medical reports together with

the corresponding consumer terms (even though

they need further verification) and provide the

related definitions. Of course, more experiments are

needed and we are in the process of executing them

with a group of physicians and patients.

Notice that, as said above, our system does not

create any change in the original text by replacing

the word or inserting an explanation in the text

because, in our opinion, this could disorient the user.

It only provides a translation (as a tooltip) and

additional info (on a separated frame) on request,

leaving the user fully in charge of his/her navigation

path through the text as it was originally created.

5 CONCLUSIONS AND FUTURE

WORK

In this paper we have presented a system that, given

a medical text, automatically finds the technical

terms, translates them in plain language and

provides additional information for them with the

same kind of language. We have implemented the

system and built a first prototype that is working and

provides satisfying results but needs some

improvements.

As a first priority, the medical vocabulary needs

to be expanded to be able to find more technical

terms present in the text. Very important is the

completion of the Italian CHV by adding more

HEALTHINF2015-InternationalConferenceonHealthInformatics

626

technical terms and their consumer equivalents.

Finally, other health-consumer dictionaries have to

be found for increasing the number of definitions

that come from medical sources.

A potential extension of the system comes from

providing the user with direct access to an external

web search engine, either a generic one (i.e. Google

or Bing) or a specific one (such Quertle or Pubmed),

for finding further information beside the ones

already provided. Also, consumer-oriented sites

could be directly accessed from the system so to

allow the user an easy navigation inside familiar

information.

We plan to complete the prototype and integrate

it within an Electronic Health Record (EHR) or

Personal Health Record (PHR) by using the HL7

“Infobutton” standard. Moreover, we want to extend

it so that it is able to take as an input the URL of any

health web page and automatically provide, as

output, the same page with the highlighted technical

terms together with their consumer translations and

definitions so to help a generic user in understanding

any health-related web page.

ACKNOWLEDGEMENTS

This work was partially funded by the PON Smart

Cities PON04a2_C “SMART HEALTH –

CLUSTER OSDH – SMART FSE-STAYWELL”

project.

REFERENCES

Baorto, D.M. & Cimino, J.J., 2000. An “infobutton” for

enabling patients to interpret on-line Pap smear

reports. AMIA Annual Symposium Procs., pp.47–50.

Cimino, J., 2007. An integrated approach to computer-

based decision support at the point of care. Trans. of

the American Clinical and Climatological Ass., 118,

pp.273–288.

Consumer Health Vocabulary Initiative. <Available from:

http://consumerhealthvocab.org/>. (6 October 2014).

Dizionario della salute. <Available from: http://

www.corriere.it/salute/dizionario/>. (6 October 2014).

Dizionario di Medicina e Biologia. <Available from:

http://medicina.zanichellipro.it/>. [6 October 2014].

HL7 Version 3 Implementation Guide: Context-Aware

Knowledge Retrieval Application (Infobutton),

Release 4. <Available from: http://

www.hl7.org/implement/standards/product_brief.cfm?

product_id=22>. (6 October 2014).

HL7 Product Infobutton. <Available from: http://wiki.hl7.

org/index.php?title=Product_Infobutton >. (6 October 2014).

Infobuttons. <Available from: http://clinfowiki.org/

wiki/index.php/Infobuttons>. (6 October 2014).

Italian Consumer-oriented Medical Vocabulary.

<Available from: https://ehealth.fbk.eu/resources/

italian-consumer-oriented-medical-vocabulary-icmv>.

(6 October 2014).

Kandula, S., Curtis, D. & Zeng-Treitler, Q., 2010. A

semantic and syntactic text simplification tool for

health content. AMIA Annual Symposium Procs.,

pp.366–370.

Kemper, D. et al., 2010. Getting patients to meaningful

use: using the HL7 infobutton standard for information

prescriptions. Healthwise White Paper Series.

Keselman, A. & Slaughter, L., 2007. Towards consumer-

friendly PHRs: patients’ experience with reviewing

their health records. AMIA Annual Symposium Procs.,

pp.399–403.

Keselman, A., Smith, C. A., 2008 et al.. “Consumer health

concepts that do not map to the UMLS: where do they

fit?”, Journ. Am. Med. Inform. Ass., vol. 15, no. 4, pp.

496-505.

Levenshtein Distance Algorithm. <Available from:

http://en.wikipedia.org/wiki/Levenshtein_distance>.

(6 October 2014).

MedlinePlus. <Available from: http://www.nlm.nih.gov/

medlineplus/healthtopics.html>. (6 October 2014).

Ok Salute. <Available from: http://www.ok-

salute.it/dizionario-medico>. (6 October 2014).

Plain Language.gov. <Available from:

http://www.plainlanguage.gov/>. (6 October 2014).

Seedorff, M. & Peterson, K., 2013. Incorporating Expert

Terminology and Disease Risk Factors into Consumer

Health Vocabularies. Pacific Symposium in

Biocomputing, pp.421–432.

Simple English Wikipedia. <Available from:

https://en.wikipedia.org/wiki/Simple_English_Wikipe

dia>. (6 October 2014).

Unified Medical Language System. <Available from:

http://www.nlm.nih.gov/research/umls/>. (6 October

2014).

WebMD. <Available from: http://www.webmd.com/a-to-

z-guides/common-topics/default.htm>. (6 October

2014).

Zeng, Q. & Tse, T., 2006. Exploring and developing

consumer health vocabularies. Journal of the

American Medical Informatics, pp.24–29.

Zeng-Treitler, Q. & Goryachev, S., 2007. Making Texts in

Electronic Health Records Comprehensible to

Consumers : A Prototype Translator. AMIA Annual

Symposium, pp.846–850.

Zielstorff, R.D., 2003. Controlled vocabularies for

consumer health. Journal of Biomedical Informatics,

36(4-5), pp.326–333.

AnAutomaticSystemforHelpingHealthConsumerstoUnderstandMedicalTexts

627