=Paper=
{{Paper
|id=Vol-3070/paper09
|storemode=property
|title=Cardiovascular Disease Risk Predictor Using ISO 9241-210 and Artificial Intelligence Techniques: A Case Study
|pdfUrl=https://ceur-ws.org/Vol-3070/paper09.pdf
|volume=Vol-3070
|authors=Misael Zambrano-de la Torre,Huizilopoztli Luna-García,Carlos E. Galván Tejada,Jorge I. Galván-Tejada,Hamurabi Gamboa-Rosales,José M. Celaya-Padilla,Maximiliano Guzmán-Fernández,J. Guadalupe Lara-Cisneros,Luis A. Flores-Chaires,Miguel A. Fraire-Hernandez
}}
==Cardiovascular Disease Risk Predictor Using ISO 9241-210 and Artificial Intelligence Techniques: A Case Study ==
https://ceur-ws.org/Vol-3070/paper09.pdf
Cardiovascular Disease Risk Predictor Using ISO 9241-
210 and Artificial Intelligence Techniques: A Case Study
Misael Zambrano-de la Torre1, Huizilopoztli Luna-García1, Carlos E. Galván
Tejada1, Jorge I. Galván-Tejada1, Hamurabi Gamboa-Rosales1, José M. Celaya-
Padilla1, Maximiliano Guzmán-Fernández1, J. Guadalupe Lara-Cisneros1, Luis A.
Flores-Chaires1, Miguel A. Fraire-Hernandez1
1 Unidad Académica de Ingeniería Eléctrica, Universidad Autónoma de Zacatecas, Jardín
Juárez #147, Centro Histórico C.P. 98000, Zacatecas, México.
{zambranot1, hlugar, ericgalvan, gatejo, hamurabigr, jose.celaya, jglara, luischaires,
miguel.fraire}@uaz.edu.mx, maxguzman1@hotmail.com
Abstract. Currently, people suffer different diseases that degenerate their life
quality. In Mexico, the number of deaths per year is commonly caused by
cardiovascular diseases. Frequently, patients ignore having this condition until
their health is complicated. For this reason, this article presents the design of a
mobile application prototype of low fidelity, based on User-Centered Design.
Using artificial intelligence techniques, data is collected to determine the
patient's cardiovascular condition. In other words, the patient is classified early
about the risk of cardiovascular disease. By joining UCD and ML. the first
prototype of the cardiovascular disease risk predictor was found. The User-
Centered Design (UCD) stages are implemented based on the ISO 9241-210:
2019 standard. The contribution of this work is the implementation of a simple
and easy diagnostic tool. The evaluation and validation of the prototype was
done through focus groups to carry out satisfaction and usability tests, obtaining
satisfactory results.
Keywords: Cardiovascular disease, Mobile technology, User-Centered Design
(UCD), Prototype, AI, K-Nearest Neighbors, Random Forest.
1 Introduction
Recently, according to the World Health Organization (WHO), degenerative diseases
kill approximately 41 million people per year, accounting for 71% of global deaths
[1]. Among the main degenerative diseases there are cardiovascular diseases. These
diseases cause the death of 17.9 million people worldwide. In Mexico, in 2019,
through reports from the Forensic Medical Services, Civil Registry and other
institutions, a total of 747, 784 deaths were registered [2]. Where cardiovascular
diseases occupy 23.5%, approximately 156,041 deaths. Based on the previous, an
analysis is made in the development of prototypes for the prevention of
Copyright © 2021 for this paper by its authors.
Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
cardiovascular diseases. Tools such as the Machine learning (ML), User-Centered
Design (UCD) and mobile technologies create solutions to this type of problem. ML
and UCD are used in the development of products and systems in healthcare [3], e.g.,
The use of UCD for diabetes detection. Where the UCD methodology was used to
develop a mobile application for the care and prevention of diabetes in Mexico [4].
This research shows the different methodologies to be followed to develop user-
centered products for health. All this based on UCD [5].
One way to enable the interaction between the human (user) and the ML is through
mobile technology. There are mobile applications that function as a tool for patient
healthcare. Such as administrative tasks like a medical consultations and services of
the Mexican Institute of Social Security (IMSS) [6]. This type of advances in
technology that involve mobile applications and UCD. can be useful in the work of
healthcare experts and the Mexican healthcare system in general. Therefore, artificial
intelligence is necessary for the development of supervised prediction algorithms
focused on the health area. This allows the creation of user-centered tools (UCD) and
also allows for competitive and simple technology in healthcare (ML) [7].
Machine learning (ML) analyzes different medical data and finds their correlation.
This allows it to generate predictions about medical problems and scenarios. This
reason it is important in the development of efficient and accessible medical care for
patients [8]. By using artificial intelligence, people who are at risk of developing heart
disease are studied and classified [9]. Currently there are many applications designed
for people who seek to control and monitor factors related to heart disease, for
example, blood pressure, heart rate, and others. These types of applications are
created by programmers without the knowledge of a healthcare professional. As
mentioned by David C. Klonoff, in mHealth for diabetes [11].
It is for this reason that there are different techniques that can be used for the
design of a mobile application. But there is a specific methodology that is based on
gathering and satisfying the needs of the target user, called User-Centered Design
(UCD). It is a methodology created for designers to adapt their products to the needs
of the target users and not the other way around. The objective of User-Centered
Design is to understand the requirements of the target user. To succeed in developing
useful, attractive and efficient systems for the users. In this way, to take advantage of
the requirements provided and collected from the target user [12].
Derived from the mentioned above, this work develops the design of a prototype of
a mobile application, which can be implemented in conjunction with an artificial
intelligence technology. Using the methodology and stages of User-Centered Design
(UCD), according to ISO 9241-210: 2019 [13]. This work provides a first approach to
a tool for the prevention of cardiovascular disease risk. Through the processing of
data obtained from users and taking the Risk Factor Questionnaire [14] as a reference.
The user is provided with a quick and reliable prediction of the risk of cardiovascular
disease. Recommendations for cardiovascular health care are suggested and an alert is
sent to the physician or security contact in case of high risk or evident danger.
The main objective of the work is to provide a first prototype of a cardiovascular
risk predictor that combines the advantages of DCU design and ML technology. A
tool entirely focused on cardiovascular care. In addition, to open a breach in the
design of applications focused on degenerative diseases that implement different
technologies such as ML (this work) or add sensors and actuators (future work).
2 Related Works
It is known that different technologies have been implemented in the healthcare area,
specifically talking about sensors and devices with UCD. For example, the
stimulation of patients suffering from dementia [15], remote monitoring of patients
with chronic diseases [16] and healthcare systems to support medical staff and
confront the constant demand for medical services [17].
There are already different mobile applications based on UCD, that help users with
their health care. However, in this paper, the purpose is not to compare them. It is to
give an idea and context of what exists in the market, for example, application
designed based on UCD for the care and treatment of diabetes. [4]. Another clear
example is Everhealthier Women, a mobile health application, designed to provide
women with an easy access to preventive health information [18]. There are also
applications focused on cardiovascular disease care. Blood Pressure Watch, allows
observing graphs related to the user's blood pressure statistics and sharing them with
the healthcare professionals. In addition, it provides feedback to the user with
reminders to record their blood pressure daily [19].
On the other hand, there is also a huge number of applications in other areas that
collect specific user requirements [7]. All these applications have usability as a
fundamental principle, and therefore, the target user as their main objective.
The prototype proposed in this work differs from the above, as it includes other
specific characteristics focused entirely on the patient at risk of cardiovascular
disease. Some of them are: age, sex, chest pain and others. The prototype allows
feedback to the user with a clear prediction of suffering a cardiovascular disease. In
the case of obtaining a high level of risk, immediately contact healthcare professional,
as well as notifying a trusted contact. In addition, new features, such as serum
cholesterol and glucose level, can be incorporated into the final design of the
application. Finally, preliminary results of implementing machine learning algorithms
such as Random Forest (RF) and K-Nearest Neighbors (KNN) are presented. This is
to diagnose and prevent the development of cardiovascular diseases in an early and
timely way. It is the first approximation of a prototype application that is designed
based on UCD and ML technology, as a predictor of cardiovascular disease risk.
3 Materials and Methods
User-Centered Design (ISO 9241-210) is a process that involves the target users in the
design and development stages of the prototype or product. This allows the prototype
or product to meet the requirements of the target users. In this way, preventing
complications and confusion when using the product. The International Organization
for Standardization defined ISO 9241-210: 2019, to provide recommendations for
design, based on different stages shown in Fig. 1.
Fig. 1. User-Centered Design Stages - ISO 9241-210: 2019 [13].
UCD stages according to ISO 9241-210: 2019 [13]:
1. Specification Context of Use: This is the stage in which the target users of the
system, the conditions and the application context will be identified.
Implementing the "personas" technique.
2. Specification of Requirements: Identification of the target user's requirements
and objectives, developing the system requirements based on the IEEE-830
standard.
3. Solution Production or Prototyping: Based on the data obtained in the previous
stage, possible system designs and prototypes are produced.
4. Prototype Evaluation: In this stage the prototypes are evaluated to see if they
satisfy the previous requirements.
Each of the stages observed in Fig. 1 are exemplified and developed below. For the
development of the prototype application. The different stages are separated into
the following subsections.
3.1 Stage 1: Specification Context of Use
The first step according to ISO 9241-210: 2019, is to identify the context and
requirements of the target user. In addition, determine the feedback goals to be
achieved with the development of the application. Potential target users were
interviewed to identify their knowledge about health applications. In addition, using
the "Personas" technique, a general profile of the target user was generated [20].
Thanks to this technique, the target user profile is between 30 and 65 years of age.
People who are beginning to worry about their health. They commonly already use
some type of application related to health, but not specifically to the heart.
Due to the current SARS-CoV-2 pandemic situation in the country, the
questionnaires were carried out remotely. Digital tools such as Google Forms were
used. Because it allows to create interviews for free, storing the information in graphs
for later analysis [21].
3.2 Stage 2: Specification of Requirements
Based on the information obtained in the previous stage, the target user for the
application is identified, as well as the age at which users begin to worry about their
cardiovascular health. The different features, questions and aspects of the prototype
application are based on the Risk Factor Questionnaire [14]. The main objective of
this questionnaire is to identify people with different degenerative diseases. Basic
information was collected for the development of the prototype.
3.3 Stage 3: Solution Production or Prototyping
This stage uses the information gathered in the previous two stages to generate a low-
fidelity prototype, which means that the prototypes created do not contain a definitive
appearance of the application. It serves to obtain information about the interaction of
the target user and the application [22]. To achieve the development of the application
prototype, Balsamiq Mockups 31 [23] was used. It is a low-fidelity interface design
tool that uses wireframes. It allows creating code-free mockups, modifying and
reorganizing graphical elements very easily.
3.4 Stage 4: Prototype Evaluation
After the previous stages were completed, the evaluation of the prototypes was carried
out. The technique to achieve the evaluation is known as "focus groups" [24]. In
addition, a questionnaire was implemented to evaluate the user experience, user
preferences and personal opinion of the prototype [25]. To give functionality to the
prototype, Justinmind2 software was used [26]. This software allows to give
functionality to the prototypes without the need to write code of any programming
language. As a result, a working prototype was obtained and shown to a group of
people. According to Jakob Nilsen, considered as one of the fathers of usability, he
mentions that the number of users needed to perform the evaluation of a software is
from 3 to 5. According to his study, the same results are obtained with a small group
of users as with a bigger group [27]. That is why the same criterion is applied in this
work. Evidently basing the questions and tasks on health-related information.
Machine Learning Algorithms
Machine learning consists of creating models that are able to perform specific
tasks. Depending on the task, a machine learning algorithm can be chosen and
adapted to classify or predict a result.
On the other hand, there are two types of algorithm groups: supervised and
unsupervised. In this work, only supervised algorithms will be used. The group of
supervised learning algorithms is characterized by learning from the inputs and
1 Balsamiq Mockups 3, available: https://balsamiq.com/wireframes/desktop/docs/overview/
2 Justinmind, available: https://www.justinmind.com/
outputs of the database. The database used in this work, was the “cardiac disease
dataset” coming from the UCI Machine Learning Repository [28], the ddsm version.
The objective of implementing ML algorithms is to introduce the information from
this repository to the prototype that has been designed based on DCU.
This database has 14 features (for example: age, sex, etc.) and 303 instances
(patients). Among the features used to create the cardiovascular risk classification
model were: age, sex, chest pain, resting blood pressure, cholesterol, fasting blood
sugar, maximum reached heart rate, exercise-induced angina, and finally condition,
high risk (1) and low risk (0). First, a preprocessing stage was carried out. Different
null values were detected and eliminated. Subsequently, attributes were introduced to
develop two algorithms. All features were used. The Random Forest and K-Nearest
Neighbors algorithms were developed. This is because they are one of the algorithms
commonly used in the health area [29]. The development method was as follows:
75% for the training stage and 25% for the algorithm testing stage.
In order to validate the correct performance of the algorithms (how well they
classify), it is necessary to perform an evaluation of the two algorithms. To evaluate
performance, two parameters were considered: accuracy (Acc) and area under the
curve (AUC). These parameters help to determine the ability of the algorithm to
correctly classify patients at risk for cardiovascular disease or not. Accuracy can be
calculated from the values of true positives (VP, patients with correctly classified
risk), true negatives (VN, patients without correctly classified risk), false positives
(FP, patients without incorrectly classified risk), false negatives (FN, patients with
incorrectly classified risk). The accuracy value is between 0 and 1, i.e., the value close
to 1 means a better performance for the correct classification of patients.
To justify this performance, the area under the curve is calculated from the
sensitivity and specificity values. These represent the probability of the algorithm to
correctly classify patients. The area is limited between the range of 0.5 and 1. Where
if a value of 1 is obtained, it means that the algorithm perfectly classifies all patients.
On the other hand, if a value of 0.5 is obtained, it means that the algorithm is unable
to classify the patients correctly. In this way, a prototype application based on DCU
and the implementation of artificial intelligence algorithms, were combined.
Finally, following each of the stages, a low-fidelity prototype was developed to
classify whether a patient has a high or low risk of cardiovascular disease.
4 Results
From the implementation of the User-Centered Design (UCD) stages in the
application prototype, the results of each stage are shown. In addition, the
performance of the algorithms based on accuracy and area under the curve, is shown.
4.1 Specification Context of Use and Requirements
According to the questionnaire performed through Google Forms and the different
questions asked, the opinions of 35 people were obtained. The number of persons
interviewed was not higher due to the SARS-COV 2 contingency lived in the country.
The 35 persons interviewed accessed the interview through a mobile device. This
situation made it very difficult to interact with the interviewees. An age range of 19 to
61 years was obtained. Fig. 2 shows the age ranges of the people interviewed. This
figure shows that the largest number of respondents is in the 45 to 58 age range, with
a total of 24 people, i.e., approximately 68%. This is possibly due to the evident
concern of adults about the risk of suffering cardiovascular disease.
Fig. 2. Age of persons interviewed.
The main reason for users to use an application of this type is due to their interest
in taking care of their cardiovascular health and improving their lifestyle. It is evident
that a high percentage of those interviewed would have a cardiovascular problem.
Most of them are over 45 years old. Cardiovascular diseases are more frequent in this
age range, according to health experts. Based on the data obtained in the interviews,
66% of those interviewed said they already had a problem related to their
cardiovascular health. This problem is shown in Fig. 3.
Fig. 3. Persons with cardiovascular problems.
Based in the Fig. 3, a question related to different aspects to be considered in the
design of the application was included. This question is related to including an
emergency button. Where 100% of the interviewees showed an interest in including a
way to immediately contact a trusted contact or medical service. This button refers to
having immediate access to send an alert with a single click. Avoiding searching in
the phone book, the number of the trusted contact or medical service. Simply press a
button and automatically send an alert that there is a problem.
Subsequently, a question was asked about how to provide feedback to the user.
This refers to the way of displaying risk diagnosis of cardiovascular disease. Three
different forms of feedback were proposed: by percentage, by risk ranges and by
specific answer (Yes/No). 86.6% of respondents preferred to obtain concrete
feedback. Preferred to receive feedback with: “Yes” or “No”. Fig. 4 shows the
percentages for each way of displaying feedback.
Fig. 4. The way of displaying risk diagnosis of cardiovascular disease.
Based on the data obtained from Specifications Context of Use stage. The
requirements were compiled. This was obtained based on the IEEE-830 standard
[29].
Then, the low-fidelity application prototype was created. This helped to design it in
a simple way and in little time, this allows the evaluation and redesign. This prototype
does not show a definitive version of the application; however, it will help to evaluate
the functionality. The focus group method was implemented to support the user
requirement specifications.
4.2 Solution Production or Prototyping
A first design prototype was made with basic elements. But evidently it needed to be
redesigned. The colors in this first version were not very contrasting and it was
deficient to have a registry of users. This allowed feedback from the users to improve
the prototype. Two important aspects were emphasized: the registration of a user and
the way in which data such as sex, age and chest pain were entered to perform the
prediction. In addition, the colors were adapted and considered based on the
recommendations. A new redesign was implemented.
In the second version of the prototype different modifications were made. Now the
user can easily register. In this section, data to generate the output, such as age and
gender, were recorded. This reduced the number of keystrokes (clicks on the screen)
required to enter the information. This allows for easier interaction between the user
and the prototype. In this new design an improved login was included. The new login
design is shown in Fig. 5.
Fig. 5. a) Login Screen First version b) Login Screen Second version c) User Registration
Screen Second version
In the first version, user registration screen was not available. In the second version
a screen for user registration was implemented. This screen allows the user to enter
data such as user name, age, gender, and others. At the end of the registration, the user
is asked to enter directly to the parameter registration to start the classification
diagnosis.
Subsequently, the interface where the type of chest pain suffered by the user is
entered. In addition, the set of predictions previously made. A loading screen was
added. Its purpose is to provide feedback to the user on the percentage complete of
the diagnosis, as shown in Fig. 6.
Fig. 6. a) Chest Pain screen with past diagnoses b) Load screen.
At the chest pain screen, all the user's information is displayed. To enter the user's
chest pain level, just swipe the indicator. There are flags to help the user to decide the
level of pain. Once the chest pain is entered, the user needs to press the "Send Info"
button to start the diagnosis. Once the diagnosis is started, the loading screen appears.
A percentage display allows the user to view the progress of the diagnostics.
Subsequently, an interface is displayed with the diagnosis corresponding to a high
or low risk of suffering a cardiovascular disease. It is possible to observe that, in the
case of high risk, an icon is displayed to send an alert message to the medical service.
In addition, the user has the possibility to store or share the diagnosis through
different options.
In this way, the user has the possibility to store or share the diagnosis through
different options. For example, a screenshot or a .txt file. In the first case, the user
could store as an image, the diagnostic obtained, with date and time. In the case of the
.txt file, it would store each of the parameters obtained, as well as the patient's
personal information to send by email or store it in the cloud.
On the other hand, the user can generate a new diagnosis or access the telephone
menu to make a call. The diagnosis already generated will be automatically saved in
the application so that it can be consulted later in the "Latest predictions" section.
Finally, some health recommendations for the users are shown. Each
recommendation was proposed by cardiovascular health experts. This allows to offer
a feedback focused on the diagnosis obtained, as shown in Fig. 7.
Fig. 7. a) High Risk screen b) Low Risk screen c) Recommendations.
4.3 Prototype Evaluation
From the second prototype previously generated, it was tested on 5 users. Each of
them of different ages, physical condition and gender. This according to Jakob Nilsen
[26]. It was explained to the users that it was a prototype and that the result obtained
was not real, because no measurements or calculations were taken, as shown in Fig. 8.
Fig. 8. Prototype evaluation.
For the evaluation of the prototype, the focus group technique was used. The
participants were adults between 49 and 60 years of age, who were subjected to
simple usability tests. In addition, a series of questions were asked to evaluate their
satisfaction. Users were instructed to perform 6 different tasks. This was in order to
evaluate the usability over time of use. The name of the six tasks and the time that
took to user number one were: User Registration (28s), Login (13s), Info Entry (8s),
Diagnostic recognition (5s), Save diagnosis (12) and Start a new diagnosis (4s).
It can be seen that the user was able to correctly complete all tasks. The average
time to complete a diagnosis is approximately 1 minute. The task that takes the
longest time is the user registration. In other words, when using the application for the
first time the user will have to spend a little more than one minute, but this time will
be reduced later, because the user will not be required to register again. The same type
of evaluation was performed on four other users. The data for the 5 users are
summarized in Table 2.
Table 2. Performance of the 5 users and average time per task.
Average Time Per Task
No. Tasks (Minutes) Average Usage
User
Completed Time (Minutes)
T1 T2 T3 T4 T5 T6
1 6 28 13 8 5 12 4 12
2 6 25 11 9 4 13 6 11
3 6 28 12 7 6 16 9 13
4 6 26 11 7 6 15 8 12
5 6 26 14 8 5 12 10 13
From Table 2, it can be seen that the task that took users the longest time on
average to perform was task 1. This task refers to user registration that ranges from 25
to 28 seconds. The task that took the least time for users to perform was task 4. This
task refers to the recognition of the diagnosis. Users found the application intuitive
and easy to interpret. After the task analysis, a series of questions were asked. These
questions are designed to collect information about the user's experience according
with [24]. The purpose is to get a feel for the user experience and to improve the
prototype in the future.
1. How old are you?
2. On a scale of 1 to 5 (5 being the most difficult), how complicated is using
the application?
3. On a scale of 1 to 5 (5 being the highest), how much do you agree that the
application meets the objective?
4. What did you enjoy most about the application?
5. In general, what would you improve the application?
These questions were asked to the same 5 users. The following results were
obtained: For question 1, the age range was between 49 and 60 years old, where 60%
of the respondents are 54 years old. For question 2, the results were between 1 and 3,
where 1 was easy and 5 was difficult. Users expressed that the application was
intuitive and easy to operate. With respect to question 3, the results were 5 with
100%; the users stated that the prototype fulfilled the objective of predicting a patient
with a high or low risk of cardiovascular disease, in addition to providing the
necessary recommendations for taking care of their health. Users expressed a high
interest in the application. They indicated that it will help to obtain an early diagnosis
of cardiovascular problems. Users expressed a high interest in the application. They
indicated that it will help to obtain an early diagnosis of cardiovascular problems. In
addition, it is an accessible and quick guide to personal care. This helps to encourage
taking expert recommendations. Regarding improvements to the prototype, users
proposed the following: generate a more extensive history with other health-related
aspects, e.g., alarms to take medication or the specific type of disease that may
develop if care measures are not taken. In general, users demonstrated an ability to
operate and navigate the application easily.
Finally, two machine learning algorithms were trained and tested.
4.4 Evaluation of Machine Learning Algorithms
Finally, two machine learning algorithms were trained and tested. The purpose of the
development of these two algorithms was to compare which one classifies patients
better. The classification was achieved thanks to the division of the 298 patients
available in the database from the repository. Already as the null values removed in
the preprocessing stage. 75% of patients were used for algorithm training. 25% to test.
The main intention is to implement one of the two in the final version of the
prototype. This will complement the development done during the user-centered
design process. To evaluate the classification algorithms, the results obtained for
ACC, AUC and some more details are show in the Fig. 9.
Fig. 9. Algorithm’s evaluation.
From Fig. 9 it can be seen that the Random Forest algorithm classifies better (at the
top). Of the 25% of patients used for testing this algorithm, 90% are correctly
classified, according to the AUC obtained from the ROC curve. All this with an
accuracy of 86% and precision of 91%. In simple terms, if the prototype application is
used by 10 people, 9 of them will receive a correct diagnosis about the risk of
cardiovascular disease. On the other hand, the KNN algorithm is shown to be more
erroneous and to give an incorrect diagnosis (at the bottom). Therefore, it is discarded
for implementation in a final version of the prototype. It is important to note that the
prediction obtained in this section is a first result, it is clear that the algorithms can be
improved and obtain better performance.
5 Discussions and Conclusions
Given that the questionnaire used for cardiovascular diseases and complications has
been validated by the Mexican government, it was used as the basis for the creation of
the prototype application. The database used in the study does not include
demographic data from Mexico. On the other hand, it gives a clear idea that, if a
database with patients from specific areas is obtained and used, the prototype works
and adapts to the environment of use. Thanks to the User-Centered Design, it was
possible to generate a first useful approximation of an application prototype that
complies with the requirements and features of the target user. Based on the
evaluation tests and the feedback obtained from the users, the prototype has good
acceptance in functionality and design. In another words, the results obtained, allow
the work to take into consideration the use of artificial intelligence and the use of
electronic sensors. It is important to highlight that more studies and analysis are
required to improve this tool. This work allows to open the gap to start developing
applications or tools about specific health diseases. This in order to combat or detect
early problems in patients. The future goal is to develop a complete tool, i. e., more
information, features, software, testes and UX experiences. This will help to generate
a simple and easy-to-use cardiovascular disease prevention tool. Finally, confront the
problems related to deaths due to the health problems in Mexico. It is important to
note that this work shows a way to unify UCD and technology as ML. In the future
we will seek to integrate sensors and actuators capable of measuring different patient
attributes. To finally open a breach of research related to health, DCU, ML and
hardware.
Acknowledgement to CONACYT for the scholarships.
References
1. World Health Organization, “Noncommunicable diseases,” 2019. https://www.who.int/news-
room/fact-sheets/detail/noncommunicable-diseases (accessed Dec. 03, 2020).
2. World Health Organization, “Cardiovascular diseases,” 2019. https://www.who.int/health-
topics/cardiovascular-diseases/#tab=tab_1 (accessed Dec. 03, 2020).
3. P. Kshirsagar, A. Pote, K. K. Paliwal, V. Hendre, P. Chippalkatti, and N. Dhabekar, “A
Review on IOT Based Health Care Monitoring System,” in Lecture Notes in Electrical
Engineering, 2020, vol. 570, pp. 95–100, doi: 10.1007/978-981-13-8715-9_12
4. M. H. Guerrero-Flores et al., “Prevention of diabetes mellitus through the use of mobile
technology (mHealth): Case study,” in Communications in Computer and Information
Science, Jun. 2019, vol. 1114 CCIS, pp. 299–313, doi: 10.1007/978-3-030-37386-3_22.
5. M. F. Bause, H. Forbes, F. Nickpour, and D. Schaefer, “Towards a Health 4.0 Framework for
the Design of Wearables: Leveraging Human-Centered and Robust Design,” in Procedia
CIRP, Jan. 2020, vol. 91, pp. 639–645, doi: 10.1016/j.procir.2020.02.222.6.
6. IMMS, “IMSS Digital App,” 2019. http://www.imss.gob.mx/imssdigital (accessed Dec. 03,
2020).
7. S. Uddin, A. Khan, M. E. Hossain, and M. A. Moni, “Comparing different supervised
machine learning algorithms for disease prediction,” BMC Med. Inform. Decis. Mak., vol.
19, no. 1, p. 281, 2019, doi: 10.1186/s12911-019-1004-8
8. A. N. Ramesh, C. Kambhampati, J. R. T. Monson, and P. J. Drew, “Artificial intelligence in
medicine,” Annals of the Royal College of Surgeons of England, vol. 86, no. 5. Royal
College of Surgeons of England, pp. 334–338, Sep. 2004, doi: 10.1308/147870804290
9. M. Diwakar, A. Tripathi, K. Joshi, M. Memoria, P. Singh, and N. kumar, “Latest trends on
heart disease prediction using machine learning and image fusion,” Mater. Today Proc., Oct.
2020, doi: 10.1016/j.matpr.2020.09.078
10. M. H. Forouzanfar et al., “Global, regional, and national comparative risk assessment of 79
behavioural, environmental and occupational, and metabolic risks or clusters of risks, 1990–
2015: a systematic analysis for the Global Burden of Disease Study 2015,” Lancet, vol. 388,
no. 10053, pp. 1659–1724, Oct. 2016, doi: 10.1016/S0140-6736(16)31679-8.
11. D. C. Klonoff, “The current status of mHealth for diabetes: Will it be the next big thing?,”
Journal of Diabetes Science and Technology, vol. 7, no. 3. SAGE Publications Inc., pp.
749–758, May 01, 2013, doi: 10.1177/193229681300700321
12. K. Torning and H. Oinas-Kukkonen, “Persuasive system design: State of the art and future
directions,” in ACM International Conference Proceeding Series, 2009, vol. 350, p. 1, doi:
10.1145/1541948.1541989
13. ISO 9241-210, “Ergonomics of human-system interaction — Part 210: Human-centred
design for interactive systems,” International Standard, 2019.
https://www.iso.org/standard/77520.html (accessed Dec. 05, 2020)
14. Secretary of Health, “Risk Factors Questionnaire,” 2015.
https://www.gob.mx/salud/documentos/cuestionario-de-factores-de-riesgo (accessed Nov.
05, 2020).
15. E. Boumpa, A. Gkogkidis, I. Charalampou, A. Ntaliani, A. Kakarountas, and V. Kokkinos,
“An Acoustic-Based Smart Home System for People Suffering from Dementia,”
Technologies, vol. 7, no. 1, p. 29, Mar. 2019, doi: 10.3390/technologies7010029
16. M. Donati, A. Celli, A. Ruiu, S. Saponara, and L. Fanucci, “A Telemedicine Service
System Exploiting BT/BLE Wireless Sensors for Remote Management of Chronic Patients,”
Technologies, vol. 7, no. 1, p. 13, Jan. 2019, doi: 10.3390/technologies7010013
17. I. Khayal, “A Systems Thinking Approach to Designing Clinical Models and Healthcare
Services,” Systems, vol. 7, no. 1, p. 18, Mar. 2019, doi: 10.3390/systems7010018
18. J. Reyes, Y. Washio, M. Stringer, and A. M. Teitelman, “Usability and Acceptability of
Everhealthier Women, a Mobile Application to Enhance Informed Health Choices,” JOGNN
- J. Obstet. Gynecol. Neonatal Nurs., vol. 47, no. 6, pp. 853–861, Nov. 2018, doi:
10.1016/j.jogn.2018.04.139.
19. NumbersMatter2Me Development company, “Blood Pressure (BP) Watch,” 2020.
https://play.google.com/store/apps/details?id=com.boxeelab.healthlete.bpwatch&hl=en_IN
(accessed Dec 05, 2020).
20. L. Nielsen, Personas - User Focused Design. London: Springer London, 2013
21. Google, “Google Forms: Free Online Surveys for Personal Use,” Google. 2020, Accessed:
Nov. 26, 2020. [Online]. Available: https://www.google.com/intl/en/forms/about/.
22. S. J. Iribarren, J. Wallingford, R. Schnall, and G. Demiris, “Converting and expanding
mobile support tools for tuberculosis treatment support: Design recommendations from
domain and design experts,” J. Biomed. Informatics X, vol. 5, p. 100066, Mar. 2020, doi:
10.1016/j.yjbinx.2019.100066
23. Balsamiq, “Editor Overview - Balsamiq for Desktop Documentation | Balsamiq,” Balsamiq.
https://balsamiq.com/wireframes/desktop/docs/overview/ (accessed May 05, 2021)
24. M. C. Maguire, “User-Centred Requirements handbook,” Wp5, no. July, p. 184, 1998,
[Online]. Available: https://dspace.lboro.ac.uk/dspace-
jspui/bitstream/2134/2651/1/PUB493.pdf
25. D. Biduski, E. A. Bellei, J. P. M. Rodriguez, L. A. M. Zaina, and A. C. B. De Marchi,
“Assessing long-term user experience on a mobile health application through an in-app
embedded conversation-based questionnaire,” Comput. Human Behav., vol. 104, p. 106169,
Mar. 2020, doi: 10.1016/j.chb.2019.106169.
26. JustInMind, “Free prototyping tool for web & mobile apps - Justinmind,” 2020.
https://www.justinmind.com/ (accessed Dec. 07, 2020)
27. J. Nielsen, “Usability inspection methods,” in Conference on Human Factors in Computing
Systems - Proceedings, Apr. 1994, vol. 1994-April, pp. 413–414, doi:
10.1145/259963.260531
28. Dua, D. and Graff, C, “{UCI} Machine Learning Repository,” 2017.
http://archive.ics.uci.edu/ml (accessed Dec. 04, 2020).
29. Institute of Electrical and Electronics Engineers, “Especificación de Requisitos según el
estándar de IEEE 830,” 2008. Accessed: Oct. 20, 2020. [Online]. Available:
http://www.fdi.ucm.es/profesor/gmendez/docs/Is0809/ieee830.pdf