An Ontology for Clinical Decision Support System to Predict

Female’s Fertile Period

Francisco Vaz

, Rodrigo Rocha Silva

2,3

and Jorge Bernardino

1,3

Polytechnic of Coimbra - ISEC, Rua Pedro Nunes, Coimbra, Portugal

FATEC Mogi das Cruzes, São Paulo State Technological College, Brazil

CISUC – Centre for Informatics and Systems of the University of Coimbra, Coimbra, Portugal

Keywords: Ontology-based Clinical Decision Support System, Domain Ontologies, Knowledge Engineering.

Abstract: Nowadays, many women do not fully realize what the fertile period is, as well as what it represents in their

life. The prediction of fertile period is quite complex and difficult to calculate accurately and this is

undoubtedly a problem for women’s. There are still some completely wrong ideas about reproductive

health, especially about the fertile period and the menstrual cycle. A good example of the myths that persist

is the fact that many women continue to believe that ovulation occurs precisely in the middle of their

menstrual cycle, which is not always true. Therefore, to better understand the female cycle, we proposed a

Clinical Decision Support System based on the use of an ontology. Our proposal can predict the female

fertile period, based on certain factors that allow a calculation that is more accurate improving the quality of

patient life.

1 INTRODUCTION

At a certain point in a woman's life, the fertile period

gains a lot of importance. However, it is common for

many women to not fully realize what the fertile

period is and what it represents for their life. It is

easily defined as the ideal time for the woman to

conceive, which has a durability of approximately 6

days and occurs 14 days before the last of the cycle

(Wilcox, Dunson, and Baird, 2000).

Women have begun to have more access to

information about their reproductive health, but

there are still many questions about the fertile period

and the menstrual cycle. Some of these questions

are, the factors influencing the menstrual cycle, the

best way to calculate it, among others. With this

questions, it is of great relevance to be able to unveil

all the myths and all the ideas formed based on

misconceptions and to encourage the women to

know better their body.

In this paper we propose a Clinical Decision

Support (CDS) system based in the use of an

ontology to overcome the problem of predicting

accurately the female fertile period and giving

important information to the user. This system can

work partially in place of the doctors in the

prediction of the fertile period. The CDS system

allows the user to have the possibility of sharing data

with the doctor. This CDS system has several

purposes, such as establishing a diagnosis, with

access to the frequency of symptoms, for diseases

that cause influences on the menstrual cycle or in

cases of planning a pregnancy. The CDS system we

propose reduces the frequency of physical

consultations and allows a better doctor-patient

relationship, since the doctor always has at his/her

disposal the data of the user, being able to do a

enhanced follow-up. Is evident that the system

permits access to the patient historical data and thus

the comparison between cycles. The system also

allows the woman to acquire a better knowledge

about their body and improve her quality of life. Of

course, is aimed for female genre, from puberty, in

which it has the first menstrual episode (commonly

between 11 and 16 years), until menarche and

menopause (Seeley, Stephens, and Tate, 2003). And

with regard to the prediction of the fertile period, our

system has the capacity to calculate different

durations that a cycle can have, as well as the

probability of occurrence of each factor (obesity,

stress, pregnancy, among others) that has influence

in the fertile period.

Vaz F., Silva R. and Bernardino J.

An Ontology for Clinical Decision Support System to Predict Femaleâ

Zs Fertile Period.

DOI: 10.5220/0006519102880293

In Proceedings of the 9th International Joint Conference on Knowledge Discovery, Knowledge Engineering and Knowledge Management (KEOD 2017), pages 288-293

ISBN: 978-989-758-272-1

The main objective of this work is the

construction of a system that can work without the

support of health professionals at an early stage and

that predicts the stages of the menstrual cycle for a

woman. It is important that they be guided by some

factors. The factors are the first day of the next

menstruation, the risk of pregnancy for each day of

the cycle, the recording of the days in which there

were changes in mood, weight, symptoms, the day

of ovulation, but always have medical follow-up

through data sharing that the system will allow. Is

important to use ontology to support the sharing and

reuse of formally represented knowledge, and it is

useful to define the common vocabulary in which

shared knowledge is represented.

The rest of the paper is organized as follows.

Section 2 discusses the existing work on the

ontology in general and related to Ontology Based

Clinical Decision Support systems. Section 3

discusses the proposed approach. Section 4 discusses

the proposed ontology-based system. Finally, section

5 presents the conclusions and future work.

2 BACKGROUND AND

LITERATURE REVIEW

In this section, we review some works to obtain

more information and knowledge how to create the

most correct and adequate ontology for a CDS

system.

Ontology is an information organization

technique that has received special attention in the

last years, mainly with respect to the formal

representation of knowledge. Generally, they are

created by specialists, having their structure based

on the description of concepts and of the semantic

relationships between them.

An ontology typically provides a vocabulary that

describes a domain of interest and a specification of

the meaning of terms used in the vocabulary

(Gruber, 1993). Depending on the precision of this

specification, the notion of ontology encompasses

several data and conceptual models, including, sets

of terms, classifications, thesauri, database schemas,

or fully axiomatized theories (Euzenat and Shvaiko,

2007).

Otero-Cerdeira et al. (2015) present a literature

review regarding articles on ontology matching

published in the last decade. It serves the purpose of

offering an up-to-date review of the field and

showing its evolution trends. In this paper over 1600

papers have been sorted according to a classification

framework defined by the authors. This framework

helps in identifying the distribution of the load work

in the last decade.

Giovannini et al. (2012) addressed the software

part by proposing a product centric ontology in

which concepts of products, processes and resources

are associated to functions and sustainable

manufacturing knowledge. In order to project a

knowledge-based system by simulating a sustainable

manufacturing expert, can automatically identify

change opportunities and propose alternatives based

on the existing production scenario.

For our work, the ontology has an enormous

utility because in computing area, there is still no

consensus on the semantics of the term "ontology".

An ontology is also composed of a set of formal

axioms that restrict the interpretation of concepts

and relations, making a clear and unambiguous

representation on the knowledge of the intended

domain. With that we can build this system and

create an ontology for our domain, which is female’s

fertile period.

The category we address in this paper is based on

the domain ontology, cited by (Silva, Ferreira, and

Vijaykumar, 2010) as being reusable ontologies in a

certain domain providing vocabularies on the

concepts and relationship of them within that

domain, on the elementary theories, and principles

that govern the domain and also on the activities

involved in this area.

These days, in medical settings, doctors or other

entities often make mistakes in several areas of

health. These errors can easily be avoided

considering the use of the CDS systems based on an

ontology. The CDS system help doctors daily,

increasing the quality of care provided to a patient.

These systems support doctors in the decision-

making process and suggest appropriate treatments.

The ontology is best suited to encapsulate the

concepts and relationship of terms associated with

the medical domain. It is suitable for capturing

medical knowledge in a formal way, allowing to

share and reuse it whenever necessary.

Many studies have been done with CDS systems

to help in several diseases. Manzoor, Balubaid,

Usman, and Mueen (2015) proposed a system to

predict high-risk pregnant woman. They proposed a

framework based on a CDS system that can work in

place of doctors to diagnose high-risk pregnant

women and, if this happens, refer them to qualified

doctors for the necessary treatment. In this way, high

risk patients will receive the treatment at the correct

time, and many lives can be saved. The main focus

of the system that has been proposed is to build a

diagnostic procedure that can work instead of

qualified physicians and identify high-risk patients,

once identified, can be treated by doctors. With this

system it is possible to greatly reduce the workload

of doctors, in addition to providing basic health care

to more patients (Manzoor et al., 2015).

Highlighting also the authors in (Jostinah Lam,

Abdullah, and Eko Supriyanto, 2015) that said there

is a large shortage of doctors in rural areas and that

is the main cause for maternal mortality to be quite

high.

Then a solution was proposed introducing a new

architecture of a CDS system in the field of high risk

pregnancy. The proposed architecture is composed

of seven main components. The need for CDS was

investigated through the interview session,

distribution of questionnaires and observation. The

CDS architecture was categorized into 7 major

components: knowledge base, inference engine,

machine learning, case database, EMR, query engine

and user interface (Jostinah Lam et al., 2015).

Raza, Chaundry, and Zaidi (2017) have done a

study to accurately distinguish between finger

tremors of Parkinson’s disease using a tri-axial

gyroscope. The study is an effort to provide

physicians with a CDS system to facilitate them in

accurate diagnosis of Parkinson’s disease. They

designed the hardware to acquire angular

displacement from tri-axial gyroscope and apply a

series of techniques to extract different features in

time and frequency domains. A total of 104 people

participated in their study, using resources from

these data, they were able to create a CDS system

with overall accuracy of 82.43%. They used the

CDSS in a hospital with an accuracy of 77.78%

(Raza, Chaundry, and Zaidi, 2017).

Semenov and Kopanitsa (2016) present a process

of development and implementation of a decision

support system for laboratory service patients. The

system allows patients reading and understanding

medical records in natural language. For the

laboratory service the system allowed increasing the

level of satisfaction of the patients and the number

of patients who came back to the laboratory service

for more detailed testing (Semenov and Kopanitsa,

2016).

Jabez Christopher, Khanna Nehemiah, and

Kannan (2015) presents a CDS system to assist

junior clinicians in the diagnosis of Allergic

Rhinitis. In their study, they did intradermal skin

tests were performed on patients who had plausible

allergic symptoms. For their study 872 patients who

had allergic symptoms were considered. The rule

based classification approach and the clinical test

results were used to develop and validate the CDSS

(Jabez Christopher, Khanna Nehemiah, and Kannan,

2015).

Tams and Euliano (2015) share lessons learned

from creating two respiratory CDS systems for

ventilating patients in a critical care setting. They

concluded that: when creating a CDS system you

must seek input from trained clinicians who are

willing and capable to make prompt and correct

therapies; Clinical decisions are case sensitive; CDS

system’s may not have acess to all of the data

required to make decisions, but sometimes simple

modifications to the algorithms may dramatically

improve performance; and that it is important to

focus only on the values which prove to be relevant,

because we have vast amounts of data available for a

clinician to understand the overall scope of the

patient (Tams and Euliano, 2015).

3 PROPOSED TECHNIQUE

For the construction of the CDS system we should

take in consideration several factors: the day of the

start of the last menstruation, the number of days in

the cycle, number of days of menstruation,

contraceptive methods, weight, calendar, pregnancy,

mood and symptoms status, data sharing and

notifications. With all these factors, the system

meets what is necessary to achieve the resolution to

our problem. We selected the main factors that stand

out and have more influence the female’s fertile

period. In this section, it is explained the

architecture, as well as the reason of each one of the

chosen factors, based on the study realized. And

since we understand that the question of the fertile

period is so important for women, it must be

addressed.

The proposed architecture for the CDS system is

composed of three main components, which are

inference engine (IE), knowledge base (KB) and

machine learning (ML), as shown in Figure 1. The

IE uses the knowledge on the system and the

knowledge about the patient to draw conclusions

regarding certain conditions. The IE controls what

kind of actions need to be taken by the system. IE

determines the reminders, alerts and conclusions to

be displayed in the system. The knowledge

represented by KB is used by IE and the KB may be

built with the help of an automated process, like

machine learning (ML) or field rules. In this

automated process, knowledge is acquired from

databases that have information about the users. The

KB together with ML system will complete the

whole decision-making process. The ML component

will be done in the future work to improve our

system.

Figure 1: CDS system architecture.

Figure 2 shows the overall scheme for calculate

female’s fertile period. We can see in the middle of

the figure, the central part of the system: the

database and the server. It is there where all the

communication takes place and all the information

concerning the users and the doctors is stored.

From the central part of the scheme, there is the

interaction of the two parts who can communicate in

this system, user and doctors, represented to the left

and to the right, respectively. On the left side, it was

decided to put the image of a woman to be consulted

by a doctor since, with our system a woman with a

mobile phone can communicate with the doctor.

It should be noted that it is possible for a doctor

to be aware of the condition of a patient without

having to be physically with it, just in case of need.

Thus, Figure 2 aims to demonstrate in a simple

but concrete way, the scheme for the proposed

system.

Figure 2: Overall scheme for calculate female’s fertile

period.

All the factors referenced above will now be

explained and justified. These factors will be

considered as input to the system and inserted by the

user.

 Day of the start of the last menstruation: this

corresponds to the first day of the cycle. It is

of utmost relevance for the calculation of the

fertile period. Over time, the system will be

more suitable for the user.

 Number of days in the cycle: perceive if it is a

short, long or normal cycle and allows to have

the perception of the regularity of the cycles.

Also having great relevance for the calculation

of the fertile period, because at the last day of

the cycle will be taken 14 days and will be

obtained the probability of the day on which

ovulation occurs. The mean duration of the

cycle is 22 to 36 days (Fehring, Schneider,

and Raviele, 2006).

 Number of days of menstruation: For woman to

have more knowledge about her body. The

average duration of menstruation is from 2 to

7 days (Seeley et al., 2003).

 Contraceptive method: in relation to the pill, the

existing calendar allows the user to note

whether the pill has been taken, whether it was

late or note. It should be noted that it is

possible to use any contraceptive method and

that the system will be adapted.

 Calendar: this is a very important feature of the

system, due to the fact that the ovulation day,

the days of the fertile period, the risk of

pregnancy (low, medium and high risk) for

each day of the menstrual cycle, the days of

menstruation inserted by the user and access

to the history of the previous months, are all

shown in the calendar.

 Weight: it is a factor that influences the

oscillations of the fertile period. People with

low weight do not have sufficient amounts of

fat, with this, the cycles become increasingly

irregular. Often menstruation may not even

come. Usually weight gain helps reverse the

previous scenario, which is low weight. This

condition is quite common in athletes and in

overworked women. It was found that body

weight and dissatisfaction with the body

varies during the menstrual cycle (Teixeira,

Damasceno, Dias, Lamounier, and Gardner,

2013).

 Pregnancy: there is the possibility of applying

the mode of pregnancy in the system, if the

woman gets pregnant, serving as reference and

keeping the record of each month of gestation.

The pregnancy mode allows to see the

countdown to the baby’s birth and allows to

receive a reminder to record the cycle after

pregnancy. Taking into account the doctor's

knowledge to make a more present follow-up

to the patient.

 Mood and Symptoms Status: it allows the

woman to make comparisons later between

the phase of the cycle and a particular

symptom. Turns out to be her clinical history

related to symptoms favouring the knowledge

of her body. With regard to pre-menstrual

tension it turns out to be a recurrent pattern of

emotional, physical and behavioural changes

in the days before menstruation (Seeley et al.,

2003). When they perceive that they are the

same with the cycles, it will cause the week

before menstruation more easily. The mood

and symptoms status can cause changes in the

calculation of the fertile period and for the

system it is of important relevance that one

can adapt to these factors, so that the

calculation of the fertile period is the most

reliable possible.

 Data Sharing: this allows an exchange of data

between the doctor and anyone the user

wishes to have access to. That is, the user can

always have the doctor's care without physical

presence. And regarding the sharing between

the couple, it is important that the man or

woman also follow and that has knowledge of

the body of his partner.

 Notifications: it will inform the user about

taking the pill (if applicable) and predicts the

day on which the next menstruation will

occur. It promotes the fact of the daily intake

at the same time, taking to the maximum

efficiency of the same. In the case of another

contraceptive, or even none at all, the system

will adapt it and notify, for example, only the

prediction of the day of the next menstruation.

4 PROPOSED

ONTOLOGY-BASED

SYSTEM

In this section, the ontology is discussed based on

the factors identified and justified above. Through

the above factors, an ontology capable of solving

this problem of the calculation of the female’s fertile

period was implemented.

With regard to the proposed ontology for this

domain, Figure 3, it consists of 6 main classes which

are the symptoms, mood states, contraceptive

methods, calendar, weight and pregnancy, the choice

of these six classes comes from the study carried out

in section 3.

In the symptoms status class, there are several

subclasses like headaches and abdominal cramps,

being able to have more according to the most

relevant symptoms for the calculation.

In the mood status class, it happens as in the case

of symptoms, where we have all the states that

influence the calculation.

In the class of contraceptive methods are

considered those contraceptives that currently exist.

The Calendar class has two subclasses, one of

which allows the sharing of data between users of

our system, as well as the subclass of notifications to

enable better effectiveness, for example in taking a

contraceptive method at a specific time, or warning

about the next menstruation.

We also have the pregnancy classes that serves to

indicate whether or not a woman is pregnant. This is

a very important factor for the calculations.

Finally, we have the weight class, which

according to some studies, (Teixeira et al., 2013)

proved to be a necessary factor in our ontology,

since the weight variations interfere in the

calculation of the fertile period.

Figure 3: Snapshot of female’s fertile period ontology.

In this section, a figure of great relevance was

explained (Figure 3). Ontology has the advantage

where the knowledge can be changed easily if

knowledge of the domain has been outdated. By

using ontological approach, the knowledge

repository becomes more flexible to changes.

The proposed ontology provides the controlled

vocabulary required for the annotation of our dataset

with the woman and doctor’s information,

facilitating the retrieval of and, more generally,

access to information. Such standardization also

facilitates the exchange of information and

contributes to semantic interoperability among

system.

Our proposed ontology are also critical to

hypothesis generation and knowledge discovery in a

data-driven approach to predict the delay of the

woman cycle for the next month.

5 CONCLUSIONS AND FUTURE

WORK

This work describes the importance of ontology-

based CDS systems in calculate female’s fertile

period. It also improves the quality of health care

service helping the society. The fact that this system

has the option of data sharing is an increased value,

since it allows a doctor-patient communication, quite

efficient. We must not forget that the woman only

shares what she wants, so that there is privacy for

her. It also allows to save and optimize resources,

such as the time spent on visits and consultations, as

well as optimize the diagnosis by the doctor, because

it has a history of the patient in question. The

ontological approach had been used as a foundation

in the development of CDS systems.

As future work we intend to develop a CDS

system based on the proposed architecture. This

prototype will be evaluated for its capability and

usability, and it is necessary to put into practice

using machine learning algorithms. Thus, with the

role of machine learning in the proposed CDS

system architecture, it will be possible to calculate

more accurately the female’s fertile period, and also,

for example, calculate the probability of delaying

menstruation in a particular user based on the past

data thereof. The possibility of using field rules in

this CDS system may also be studied, since it could

also be included here.

With all these implementations of the future

work made, it will be possible to do one of the most

important purposes, which are tests with women and

doctors, to prove the effectiveness of our system.

REFERENCES

Euzenat, J. and Shvaiko, P., Ontology matching. Springer,

2007

Fehring, R. J., Schneider, M., and Raviele, K. (2006).

Variability in the Phases of the Menstrual Cycle.

Journal of Obstetric, Gynecologic and Neonatal

Nursing, 35(3), 376–384.

Gruber, T. R. (1993). A translation approach to portable

ontology specifications. Knowledge Acquisition, 5(2),

199–220.

Jabez Christopher, J., Khanna Nehemiah, H., and Kannan,

A. (2015). A clinical decision support system for

diagnosis of Allergic Rhinitis based on intradermal

skin tests. Computers in Biology and Medicine, 65,

76–84.

Jostinah Lam, Abdullah, M. S., and Eko Supriyanto.

(2015). Architecture for clinical decision support

system (cdss) using high risk pregnancy ontology.

ARPN Journal of Engineering and Applied Sciences,

10, 9.

Manzoor, U., Balubaid, M. A., Usman, M., and Mueen, A.

(2015). Ontology-Based Clinical Decision Support

System for Predicting High-Risk Pregnant Woman.

International Journal of Advanced Computer Science

and Applications, 6, 6.

Raza, M. A., Chaundry, Q., and Zaidi, S. M. T. (2017).

Clinical Decision Support System For Parkinson’s

Disease and Related Movement Disorders, 1108–

1112.

Seeley, R. R., Stephens, T. D., and Tate, P. (2003).

Anatomia e Fisiologia (6

Edição).

Semenov, I., and Kopanitsa, G. (2016). Development of a

Clinical Decision Support System for the Patients of a

Laboratory Service. Studies in Health Technology and

Informatics, 228, 90–94.

Silva, R., Ferreira, M. G., and Vijaykumar, N. (2010). An

Object-Oriented Meta-Model as Ontology for

Describing Domains and Problems for Planning Space

Applications Planning. SpaceOps 2010 Conference,

(August).

Tams, C. G., and Euliano, N. R. (2015). Creating clinical

decision support systems for respiratory medicine.

2015 37th Annual International Conference of the

IEEE Engineering in Medicine and Biology Society

(EMBC), 32669, 5335–5338.

Teixeira, A. L. S., Damasceno, V. O., Dias, M. R. C.,

Lamounier, J. A., and Gardner, R. M. (2013).

Association between Different Phases of Menstrual

Cycle and Body Image Measures of Perceived Size,

Ideal Size, and Body Dissatisfaction. Perceptual and

Motor Skills, 117(3), 892–902.

Wilcox, A. J., Dunson, D., and Baird, D. D. (2000). The

timing of the “fertile window” in the menstrual cycle:

day specific estimates from a prospective study. Bmj,

321, 1259–1263.