Backpropagation Neural Network Levenberg-Marquardt Method in

Predicting Lung Cancer in Smokers

Muhammad Iqbal

, Muhammad Rafai

and Solikhun

Faculty of Science and Technology, University Pembangunan Panca Budi, Medan, Indonesia

Engineering Study Program, STIKOM Tunas Bangsa, Pemtangsiantar, Indonesia

Keywords:

Backpropagation Neural Network, Levenberg-Marquardt Method, Predicting Lung Cancer.

Abstract:

Levenberg Marquardt is a method of the Backpropagation artiﬁcial neural network algorithm. Today, many

people are infected with lung cancer. Lung cancer is one of the biggest contributors to cancer in the world.

With the development of increasingly advanced and rapid technology, people with cancer can be analyzed,

which will then be stored as data about the characteristics of people with cancer. In this study, researchers

optimized the Backpropagation artiﬁcial neural network using the Levenberg Marquardt method to predict

lung cancer in smokers. Input data used in this study are 15 variables from x1 to x15. This lung disease data

is taken from Kaggle, which consists of 309 records. The result of this study is backpropagation optimization

using the Levenberg Marquardt method to predict lung cancer in smokers with training MSE = 0.000133 and

best test = 0.00001974 with 15-6-1 architecture.

1 INTRODUCTION

Lung cancer is an abnormal condition found in the

lungs characterized by abnormal cell growth or what

is known as a dangerous tumor. Abnormal cell growth

conditions can originate from cells present in the

lungs. However, this abnormal cell growth can origi-

nate from cancer cells in other body parts that spread

to the ungs(Alsharairi, 2019). Lung cancer can occur

in both men and women(Smolle and Pichler, 2019).

Artiﬁcial Intelligence (AI) is a term that implies

using computers to model intelligent behavior with

minimal human intervention (Solikhun et al., 2020).

There are many methods in AI, one of which is Ar-

tiﬁcial Neural Network Backpropagation (Sewunetie

and Kov

acs, 2022). Backpropagation is one of the al-

gorithms in artiﬁcial neural networks that are formed

with several layers to change the weights. Changing

the weight is done with a training algorithm to get the

optimal weight (Wright, 2022). The weakness of the

Backpropagation algorithm is the poor convergence

speed and unstable learning, so it is often stuck at a

local minimum. So we need an algorithm to speed

up Backpropagation training (Manik et al., 2019; Li

et al., 2019).

The Levenberg-Marquardt algorithm is a develop-

ment of the standard backpropagation algorithm[8].

The Levenberg-Marquardt algorithm is performed

because of its convergence speed (Mikhaylov and

Tarakanov, 2020; Tan and Lim, 2019; Gavin, 2023;

Moayedi et al., 2020). Of course, by using several ar-

chitectural patterns and seeing how far the accuracy,

epoch, and times of the two algorithms are (Wisesty

et al., 2021; Liu et al., 2021) . The data used to test the

optimization of the Lavenberg-Marquardt algorithm

is Lung Cancer (Does Smoking cause Lung Cancer)

data. Data source from www.kaggle.com. Today,

many people get cancer, especially lung cancer, many

of which are infected by smoking. But not a few also

contracted lung cancer due to other factors (Andayani

et al., 2019).

So the authors conducted a lung cancer predic-

tion study using the Levenberg Marquardt Backprop-

agation artiﬁcial neural network method because it

can provide accurate results. Researchers predict the

recognition of lung cancer using 15 input variables,

namely Gender, Age, Smoke, Yellow Finger, Anxi-

ety, Peer Pressure, Chronic Disease, Fatigue, Allergy,

Wheezing, Alcohol Consuming, Coughing, Shortness

of breath, Swallowing Difﬁculty, Chest Pain. To

introduce lung cancer prediction, researchers used

100 training data and 40 test data with the Leven-

berg Marquardt method. This study tested seven net-

work architecture models, and the best architecture

was obtained. Namely, 15-6-1 with training MSE =

0.000133 and best testing MSE = 0.00001974.Artiﬁ-

Iqbal, M., Rafai, M. and Solikhun, .

Backpropagation Neural Network Levenberg-Marquardt Method in Predicting Lung Cancer in Smokers.

DOI: 10.5220/0012448200003848

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Advanced Information Scientiﬁc Development (ICAISD 2023), pages 281-284

ISBN: 978-989-758-678-1

281

cial Intelligence (AI) is a term that implies using com-

puters to model intelligent behavior with minimal hu-

man intervention. There are many methods in AI, one

of which is Artiﬁcial Neural Network Backpropaga-

tion (Choi et al., 2020).

2 RESEARCH METHODOLOGY

2.1 Methods of Data Collection

The method of data collection carried out by the au-

thor is a literature study (the sources used in the re-

search are collected from scientiﬁc journals, as well

as sources from the internet which are used for vari-

ous purposes in education).

2.2 Data Source

The data used in this study were taken from the data

website www.kagle.com, in the form of lung cancer

medical records consisting of 309 data. The data used

for training were 100 records, and the data used for

testing were 40. The following are 309 attributes in

the form of data that inﬂuence the occurrence of lung

cancer in smokers.

Table 1: Medical Record Data of Lung Cancer in Smokers.

No X1 X2 X3 .. X15 T

1 M 69 0 .. 1 Yes

2 M 74 1 .. 1 Yes

3 F 59 0 .. 1 Yes

4 M 63 1 .. 1 No

5 F 63 0 .. 0 No

6 F 75 0 .. 0 Yes

7 M 52 1 .. 1 Yes

8 F 50 1 .. 0 Yes

.. .. .. .. .. .. ..

309 M 62 0 .. 0 Yes

Input Description:

X1 = Gender

X2 = Age

X3 = Smoking

X4 = Yellow Finger

X5 = Anxiety

X6 = Social Pressure

X7 = Chronic Diseases

X8 = Fatigue

X9 = Allergies

X10 = Wheezing

X11 = Consumption of alcohol

X12 = Cough

X13 = Shortness of Breath

X14 = Hard to Swallow

X15 = Chest Pain

T = Lung Cancer

Lung cancer medical record data is converted with the

following rules:

1. Gender:

• If M(Male) then 1;

• If F(Female) then 0.

2. Age:

• Toddler Age: 0-5 years = 0.1;

• Childhood: 5-10 years = 0.2;

• Early Adolescence: 11 – 26 years = 0.3;

• Late Adolescence: 27 – 35 years = 0.4;

• Early Adulthood: 26 – 35 years = 0.5;

• Late adulthood: 36 – 45 years = 0.6;

• Early Old Age: 46 – 55 years = 0.7;

• Late Old Age: 56 – 65 years = 0.8;

• Old age: ¿ 65 years = 0.9;

3. Smoking:

• If smoking then 1;

• If don’t smoke then 0.

4. Yellow Finger::

• IIf have jaundice then 1;

• If don’t have jaundice then 0.

5. SAnxiety:

• If have anxiety disease then 1;

• If don’t have Anxiety disease then 0.

6. Peer Pressure:

• If have Peer Pressure then 1;

• If don’t have Peer Pressure then 0.

7. Chronic Disease:

• If have Chronic Disease then 1;

• If don’t have Chronic Disease then 0.

8. Fatigue:

• If have Fatigue then 1;

• If don’t have Fatigue then 0

9. Allergies:

• If have an allergic disease then 1;

• If don’t have an allergic disease then 0.

10. Wheezing:

• If have Wheezing disease then 1;

• If do not have Wheezing disease then 0.

ICAISD 2023 - International Conference on Advanced Information Scientiﬁc Development

282

11. Consumption of alcohol:

• If consuming alcohol then 1;

• If don’t consume alcohol then 0.

12. Cough:

• If have cough disease then 1;

• If don’t have cough disease then 0.

13. Shortness of Breath:

• If have shortness of breath then 1;

• If don’t have shortness of breath then 0

14. Hard to Swallow:

• If have difﬁculty swallowing then 1;

• If do not have Difﬁcult Swallowing then 0

15. Chest Pain:

• If have chest pain then 1;

• If don’t have chest pain then 0.

16. Lung Cancer:

• If YES then 1;

• If NO then 0.

Lung cancer prediction target is Lung Cancer.

That is, if infected, the value is 1; if not then the value

is 0. The results of the transformation of lung cancer

medical record data can be seen in table 2.

Table 2: Medical Conversion Data of Lung Cancer in

Smokers.

No X1 X2 X3 .. X15 T

1 1 0,9 0 .. 1 1

2 1 0,9 1 .. 1 1

3 0 0,8 0 .. 1 0

4 1 0,8 1 .. 1 0

5 0 0,8 0 .. 0 0

6 0 0,9 0 .. 0 1

7 1 0,7 1 .. 1 1

8 0 0,7 1 .. 0 1

.. .. .. .. .. .. ..

309 1 0,8 0 .. 0 1

2.2.1 Research Framework

In completing this research, the authors compiled the

research framework as follows

2.2.2 Architectural Design

The author’s architecture in this study consists of 1

input layer block, one hidden layer block, and one

output layer block. Here is an example of the 15-3-

1 architecture in use.

Figure 1: Research framework.

Figure 2: Architectural Design.

3 RESULTS AND DISCUSSION

3.1 Best Training and Testing Results

Train data and lung cancer prediction testing using the

Matlab application version R2011a with the Leven-

berg Marquardt backpropagation algorithm. The best

training and testing results are 15-6-1 with training

MSE = 0.000133 and testing MSE = 0.00001974.

3.2 Comparison of Training Results and

Testing of the Levenberg

Marquardt Method

After testing the Levenberg Marquardt method back-

propagation algorithm with architectures 15-2-1, 15-

3-1, 15-4-1, 15-5-1, 15-6-1, 15-7-1 and 15-8 -1 using

the MatLab R2011a application, the following com-

parisons can be made in table 3.

Backpropagation Neural Network Levenberg-Marquardt Method in Predicting Lung Cancer in Smokers

283

Table 3: Training and Testing.

No Architectural Epoch Performance Performance

(iterations) Testing Training

1 15-2-1 24 0.137000 0.25340000

2 15-3-1 19 0.130000 0.27500000

3 15-4-1 8 0.000597 0.00026630

4 15-5-1 8 0.000322 0.00075920

5 15-6-1 9 0.000133 0.00001974

6 15-7-1 4 0.000578 0.00036030

7 15-8-1 12 0.000302 0.00043620

4 CONCLUSIONS

Based on the results and discussion described above,

it can be concluded that the Levenberg Marquart

backpropagation method can predict potential mor-

tality in heart failure with MSE training and testing =

0.0150 with 11-7-1 architecture. Determination of the

method in backpropagation training is so inﬂuential

on the results, and it’s just that the determination of

the method and pattern must be adjusted to the needs.

REFERENCES

Alsharairi, N. (2019). The effects of dietary supple-

ments on asthma and lung cancer risk in smokers and

non-smokers: A review of the literature. Nutrients,

11(4):01,.

Andayani, U., Rahmat, R., Syahputra, M., Lubis, A., and

Siregar, B. (2019). Identiﬁcation of lung cancer using

backpropagation neural network. Journal of Physics:

Conference Series, 1361(1).

Choi, R., Coyner, A., Kalpathy-Cramer, J., Chiang, M., and

Campbell, J. (2020). Introduction to machine learn-

ing, neural networks, and deep learning. Transl. Vis.

Sci. Technol, 9(2).

Gavin, H. (2023). The levenberg-marquardt algorithm for

nonlinear least squares curve-ﬁtting problems.

Li, D., Huang, F., Yan, L., Cao, Z., Chen, J., and Ye,

Z. (2019). Landslide susceptibility prediction using

particle-warm optimized multilayer perceptron: Com-

parisons with multilayer-perceptron-only, bp neural

network, and information value models. Appl. Sci,

9(18).

Liu, F., Sekh, A., Quek, C., Ng, G., and Prasad, D.

(2021). Rs-herr: a rough set-based hebbian rule re-

duction neuro-fuzzy system. Neural Comput. Appl,

33(4):1123–1137,.

Manik, A., Adiwijaya, A., and Utama, D. (2019). Classi-

ﬁcation of electrocardiogram signals using principal

component analysis and levenberg marquardt back-

propagation for detection ventricular tachyarrhythmia.

J. Data Sci. Its Appl, 2(1):78–87,.

Mikhaylov, A. and Tarakanov, S. (2020). Development

of levenberg-marquardt theoretical approach for elec-

tric networks. Journal of Physics: Conference Series,

1515(5).

Moayedi, H., Aghel, B., Vaferi, B., Foong, L., and Bui, D.

(2020). The feasibility of levenberg–marquardt algo-

rithm combined with imperialist competitive compu-

tational method predicting drag reduction in crude oil

pipelines. J. Pet. Sci. Eng, 185.

Sewunetie, W. and Kov

acs, L. (2022). Comparison of

template-based and multilayer perceptron-based ap-

proach for automatic question generation system. In-

dones. J. Electr. Eng. Comput. Sci, 28(3):1738–1748,.

Smolle, E. and Pichler, M. (2019). Non-smoking-associated

lung cancer: A distinct entity in terms of tumor biol-

ogy, patient characteristics and impact of hereditary

cancer predisposition. Cancers, 11(2).

Solikhun, M., Saﬁi, M., and Zarlis, M. (2020). Back-

propagation network optimization using one step se-

cant (oss) algorithm. IOP Conf. Ser. Mater. Sci. Eng,

769(1).

Tan, H. and Lim, K. (2019). Review of second-order opti-

mization techniques in artiﬁcial neural networks back-

propagation. IOP Conference Series: Materials Sci-

ence and Engineering, 495(1).

Wisesty, U., Sthevanie, F., and Rismala, R. (2021). Mo-

mentum backpropagation optimization for cancer de-

tection based on dna microarray data. Int. J. Artif. In-

tell. Res, 4(2):127,.

Wright, L. (2022). Deep physical neural networks trained

with backpropagation. Nature, 601(7894):549–555,.

ICAISD 2023 - International Conference on Advanced Information Scientiﬁc Development

284