Artificial neural networks analysis of the risk factors for

aneurysm in the population of the Setif region in Algeria

BOUHARATI IMENE1,2, BOUBENDIR NASSER-DINE3, BOUHARATI KHAOULA4, LAOUAMRI

SLIMANE5

1Faculty of Medicine, Paris Sorbonne-University, FRANCE

2Laboratory of intelligent systems, UFAS Ferhat Abbas Setif University, ALGERIA,

Faculty of Medicine, Algiers University, ALGERIA

3Faculty of Medicine, Constantine University, ALGERIA

4Faculty of Medicine, UFAS Ferhat Abbas Setif University, ALGERIA,

Abstract: Background. By definition, when the abdominal aorta undergoes dilation, this is called

an aneurysm. However, this definition depends on the threshold diameter of the aorta reached after

dilation. According to angiographic studies, aneurysm is considered when the diameter of the aorta

exceeds 30 mm. According to the International Society for Cardiovascular Surgery/Society for

Vascular Surgery Ad Hoc Committee, we speak of an aneurysm when the diameter of the infrarenal

aorta exceeds 1.5 times the normal diameter. It then becomes necessary to define the normal value of

this diameter, which varies from 16 to 23 mm depending on the population concerned. Ultrasound is

often used in screening for abdominal aortic aneurysms (AAA) for its simplicity and low cost. This

study evaluates the prevalence of abdominal aortic aneurysms in the population at risk in the region of

Setif in Algeria. Method and materials. The study concerns a population of The study concerns a

population of 902 diagnosed cases, 854 cases are over 50 years old who consented to AAA screening

who consented to AAA screening. For each patient, different parameters are taken. Physical and

ultrasound examinations are performed. The parameters sex, history of diabetes, dyslipidemia, blood

pressure, body mass index, smoking and atherosclerosis are listed. In order to establish an average

diameter of the aorta in this population at risk, an intelligent analysis relating these factors to the

diameter of the abdominal aorta is applied. Conclusion. As the system is very complex to analyze

using classical mathematical techniques, the principles of artificial neural networks come in handy.

The rule base that maps the input variables to the diameter of the aorta is created from the database of

the performed analyzes. This makes it possible to predict the diameter of the abdominal aorta from the

risk factors. Therefore, prevention of abdominal aortic aneurysm will be possible in the population of

this geographic area.

Key words: Aneurysm, Abdominal aorta, Normal diameter, Aneurysm, Intelligent system, Fuzzy logic

Received: June 8, 2021. Revised: April 10, 2022. Accepted: May 8, 2022. Published: July 1, 2022.

1 Introduction

An increase in the diameter of the arterial aorta

with a loss of parallelism of its wall defines the

abdominal aneurysm [1]. There are several

definitions that are defined as the diameter

reached by the aorta. We speak of an aneurysm

in certain definitions when the diameter is

greater than 30 mm, either according to the

French Society of Vascular Medicine and the

Cardiology Society of the European Union [2].

While the International Society for

Cardiovascular Surgery/Society for Vascular

Surgery Ad hoc Committee proposes that an

AAA be defined when the maximum diameter of

the infrarenal aorta is at least 1.5 times greater

than the infrarenal aortic diameter normal kidney

[3]. As for the definition which considers the

aneurysm when the diameter of the aorta exceeds

30 mm, it fixes this threshold as being the limit

of the non-reversibility of the dilation [4-7]. This

definition does not differentiate between the

sexes. Because its value is lower in women than

in men and in patients with arteriomegaly [4].

Several factors cause the aneurysm factors such

as age where an increase is observed in the risk

of rupture and even death by up to 80% [8-9].

Generally, the risk factors are mainly smoking,

hypertension, advanced age, male sex,

WSEAS TRANSACTIONS on BIOLOGY and BIOMEDICINE

DOI: 10.37394/23208.2022.19.18

Bouharati Khaoula, Bouharati Imene,

Guenifi Wahiba, Gasmi Abdelkader,

Laouamri Slimane

E-ISSN: 2224-2902

163

Volume 19, 2022

atherosclerosis, dyslipidemia or body mass

index, and even race.

The aneurysm is considered a silent disease

without symptoms [10]. It is necessary to

establish a screening and monitoring program

for people at risk [11]. Faced with the need to

define the risk threshold according to the

definition which sets this threshold at 1.5 times

the value of the normal diameter, it is necessary

to set the normal diameter in a population at

risk. To determine this value, it is necessary to

consider all the factors involved. The system

turns out to be very complex to analyze using

classical mathematical methods. The proposal

of an intelligent system with artificial neural

networks is used to compensate for the

uncertainties inherent.

2. Material and methods

This study samples the diameter of the

abdominal aorta from 902 diagnosed patients at

the level of the radiology department of the

University Hospital of Setif in Algeria and in

the surrounding private sector clinics.

This study was spread over a period from 2018

to 2021. Clinical and physiological parameters

are taken from each patient such as sex, age,

atherosclerosis, body mass index, hypertension,

and dyslipidemia. As ultrasound remains the

most available and least invasive means, as well

as radiological technique in the detection of

vascular dysfunction [12-15], this technique was

used in our determination of the diameter of the

abdominal aorta. (Figure 1). Among the 902

diagnosed cases, 854 cases are over 50 years

old. They are therefore people at risk. For this

category, the average diameter of the aorta is

calculated (20.47 mm). As the value of this

diameter is a function of several parameters, The

values of the diameter of the aorta is considered

as output variable. The factors that determine it

are considered as input variables.

ANN system

The general principle of artificial neural

networks (ANN) is an imitation of the natural

neural network. The network comprises two

spaces (inputs-outputs).

Artificial neural networks have the dynamics and

ability to read experimental data from the real

environment and can therefore solve complex

systems of biophysical processes.

Neural networks are systems learning to perform

mapping functions between two spaces: input

space and output space.

The matching between the two spaces from the

actual measured values allows creation of a

transfer function between the two spaces. Each

time we fix the values for the input variables and

their corresponding output variable (diameter of

the aorta), the function created undergoes

readjustments. This readjustment takes place

through the modification of the mathematical

coefficients [16-18].

This is the learning phase of the network. The

network created remains the same without

changing it in each case, but these readjustments

of the function carry the adaptation out

Figure 1. Partially thrombosed fusiform

aneurysms of the infrarenal abdominal aorta. The

first nascent 50 mm below the ostium of the left

renal artery, extending 5.5 cm in height with a

maximum diameter of 40 mm.- The second S1,

extending to the bifurcation extended aorto-iliac,

about 40 mm

ANN modeling

In order to predict the average diameter of the

population of the region of Sétif in Algeria, the value

of the diameter recorded for each patient is correlated

with the clinical parameters measured. (age, sex,

BMI, blood pressure, hyperlipidemia, and

atherosclerosis) The built system allows a

correspondence function between the two spaces.

(Figure 2).

The variables sex, atherosclerosis, hypertension, and

dyslipidemia are coded (1 and 2) to express the

presence or absence. The real values express age and

body.

MATLAB 2016a is used as the compiler.

WSEAS TRANSACTIONS on BIOLOGY and BIOMEDICINE

DOI: 10.37394/23208.2022.19.18

Bouharati Khaoula, Bouharati Imene,

Guenifi Wahiba, Gasmi Abdelkader,

Laouamri Slimane

E-ISSN: 2224-2902

164

Volume 19, 2022

The mathematical expression that connects them

is of the form: f(ad) = a,g,b,ap,h,at

Where:

ad: diameter of the aorta

a: Age

g: Sex

b: BMI

ap: blood pressure

h: Hypolipidemia

at: atherosclerosis

Figure 2. Input-output system with hidden layer

Of the 854 cases whose age is greater than 50

years, half of 427 cases, their clinical parameter

variables are introduced at the entrance to the

system and the corresponding aortic diameter.

During this so-called network learning phase, it

created the transfer function. I fit this function

with the method of the fewest squares. The

gradient is 0.313. The number of adjustment

loops is 1000 (Figure 3).

Figure 3 Adjusting the function during the

learning phase

When the function is created and adjusted to its

minimum error, we inject the other half of the

remaining cases for the test phase. Note that the

test values coincide perfectly with the learning

values (Figure 4).

Figure 4. Confusion of test values with learning

Values

3. Result

The function f(x) created and adjusted with so-

called weight coefficients is presented in Figure

5. This function takes care of all the variables

and makes it possible to introduce random

values at the input to read at the output in terms

of the diameter of the corresponding aorta.

Example (Figure 5): Randomly set values

(1:55:2) refer to a male subject, 55 years old

and with atherosclerosis. The probability of

having an aneurysm is represented by a

predicted aortic diameter of 38.5 mm.

Figure 3 Quadratic function (inputs-output)

WSEAS TRANSACTIONS on BIOLOGY and BIOMEDICINE

DOI: 10.37394/23208.2022.19.18

Bouharati Khaoula, Bouharati Imene,

Guenifi Wahiba, Gasmi Abdelkader,

Laouamri Slimane

E-ISSN: 2224-2902

165

Volume 19, 2022

4. Conclusion

The diameter of the aneurysmatic

abdominal aorta is often determined by its rate

of dilation relative to its average value (1.5 the

average value). This value is specific to each

population in a geographic area. The study

carried out on the population of the region of

Setif in Algeria made it possible to match the

factors favoring aneurysm in this population.

The complexity of the system and the

interference of the effects of the factors led us to

propose a system with artificial neural networks

in the analysis of these data. The proposed

algorithm makes it possible to create a transfer

function between the two input spaces (risk

factors) and the output space (the diameter of

the aorta). We adjust this function to its

minimum error by learning the network for half

of the cases and tested by the variables of the

other half. This makes it possible to predict the

diameter of the abdominal aorta from the risk

factors. Therefore, prevention of abdominal

aortic aneurysm will be possible in the

population of this geographic area.

Financial support and sponsorship:

Nil.

Conflicts of interest

There are no conflicts of interest.

References

[1] Collège des enseignants de médecine

vasculaire et Angioweb. Anévrisme de l’aorte

abdominale. In: Traité de médecine vasculaire,

Principes de base, maladies artérielles, Tome 1.

Elsevier Masson. 2010; p.729.

[2] Starck J, Aaltonen HL, Björses K, Lundgren

F, Gottsäter A, Sonesson B, et al. A significant

correlation between body surface area and

infrarenal aortic diameter is detected in a large.

screening population with possibly clinical

implication. Int Angiol J Int Union Angiol.

2019;38(5):395-401. doi: 10.23736/S0392-

9590.19.04071-9

[3] Johnston KW, Rutherford RB, Tilson MD,

Shah DM, Hollier L, Stanley JC. Suggested

standards for reporting on arterial aneurysms.

Subcommittee on Reporting Standards for

Arterial Aneurysms, Ad Hoc Committee on

Reporting Standards, Society for Vascular Surgery

and North American Chapter, International Soci-

ety for Cardiovascular Surgery. J Vasc Surg

1991;13:452-8.3Sakalihasan N, Michel JB,

[4] Katsargyris A, et al. Abdominal aortic

aneurysms. Nat Rev Dis Primers. 2018 Oct

18;4(1):34

[5] Hirsch AT, Haskal ZJ, Hertzer NR, et al.

ACC/AHA 2005 Practice Guidelines for the

management of patients with peripheral arterial

disease (lower extremity, renal, mesenteric, and

abdominal aortic): a collaborative report from the

American Association for Vascular

Surgery/Society for Vascular Surgery, Society for

Cardiovascular Angiography and Interventions,

Society for Vascular Medicine and Biology,

Society of Interventional Radiology, and the

ACC/AHA Task Force on Practice Guidelines

(Writing Committee to Develop Guidelines for the

Management of Patients With Peripheral Arterial

Disease): endorsed by the American Association

of Cardiovascular and Pulmonary Rehabilitation;

National Heart, Lung, and Blood Institute; Society

for Vascular Nursing; TransAtlantic Inter-Society

Consensus; and Vascular Disease

Foundation. Circulation. 2006 Mar

21;113(11):e463-654

[6] Erbel R, Aboyans V, Boileau C, et al. ESC

Committee for Practice Guidelines. 2014 ESC

Guidelines on the diagnosis and treatment of

aortic diseases: Document covering acute and

chronic aortic diseases of the thoracic and

abdominal aorta of the adult. The Task Force for

the Diagnosis and Treatment of Aortic Diseases of

the European Society of Cardiology (ESC). Eur

Heart J. 2014 Nov 1;35 (41):2873-926full-text

[7] Kent KC. Clinical practice. Abdominal aortic

aneurysms. N Engl J Med. 2014 Nov

27;371(22):2101-8

[8] Ashton HA, Buxton MJ, Day ME, et al.. The

Multicentre Aneurysm Screening Study Group.

The Multicentre Aneurysm Screening Study

(MASS) into the effect of abdominal aortic

aneurysm screening on mortality in men : a

randomised controlled trial. Lancet; 2002; 360:

1531–9.

[9] Wanhainen A, Bergqvist D, Boman K, et al.

Risk factors associated with abdominal aortic

aneurysm: a population-based study with

historical and current data. J Vasc Surg. 2005; 41:

390–396.

[10] Sourabh A., Arman Q., Vishal S., and Alka

S. Abdominal aortic aneurysm: A comprehensive

WSEAS TRANSACTIONS on BIOLOGY and BIOMEDICINE

DOI: 10.37394/23208.2022.19.18

Bouharati Khaoula, Bouharati Imene,

Guenifi Wahiba, Gasmi Abdelkader,

Laouamri Slimane

E-ISSN: 2224-2902

166

Volume 19, 2022

review. Exp Clin Cardiol. Spring; 2011; 16(1):

11–15.

[11] NHS public health functions agreement

2016-17 Service specification No.23 NHS

Abdominal Aortic Aneurysm Screening

Programme.

[12] Nwafor IA, Eze JC, Ezemba N, Chinawa

JM, Idoko LF and Ngene CN. Management of

Varicose Veins of the Lower Extremities: A 10-

year Institutional Experience. J Vasc Med Surg

2017; 5: 309. doi: 10.4172/2329-6925.1000309

[13] L Rulon Hardman, J Paul Rochon. Role of

Interventional Radiologists in the Management of

Lower Extremity Venous Insufficiency.

RadioGraphics 2009; 29:525–536.

[14] KG Mercer, DJ Scott, DC Berridge.

Preoperative duplex imaging is required before

all operations for primary varicose veins. Br J

Surg 1998; 85:1495–1497

[15] RS Jutley, I Cadle, KS Cross. Preoperative

assessment of primary varicose veins: a duplex

study of venous incompetence. Eur J Vasc

Endovasc Surg 2001; 21:370–373

[16] Massard J, Ratziu V, Thabut D, Moussalli J,

Lebray P, et al. Natural history and predictors of

disease severity in chronic hepatitis C. J Hepatol.

2006. 44: S19–S24 DOI:

https://doi.org/10.1016/j.jhep.2005.11.009.

[17] Bouharati I., El-Hachmi S., Babouche F.,

Khenchouche A., Bouharati K., Bouharati S.

Radiology and management of recurrent varicose

veins: Risk factors analysis using artificial neural

networks. 2018; Volume: 5 Number:1 Pages

No:1–5]. 10.15713/ins.jmrps.116

[18] Khenchouche A., Bouharati K., Bouharati

S., Mahnane A., Hamdi-Cherif M. Post Mortem

Interval: Necrobiome Analysis Using Artificial

Neural Networks. Computational Biology and

Bioinformatics. 2017. 5(6): 90-96.

Creative Commons Attribution

License 4.0 (Attribution 4.0

International, CC BY 4.0)

This article is published under the terms of the

Creative Commons Attribution License 4.0

https://creativecommons.org/licenses/by/4.0/deed.en_

WSEAS TRANSACTIONS on BIOLOGY and BIOMEDICINE

DOI: 10.37394/23208.2022.19.18

Bouharati Khaoula, Bouharati Imene,

Guenifi Wahiba, Gasmi Abdelkader,

Laouamri Slimane

E-ISSN: 2224-2902

167

Volume 19, 2022