A MACHINE LEARNING METHODOLOGY FOR DETECTING

CHRONIC
KIDNEY DISEASE

A PROJECT REPORT

Submitted by

KAVINKUMAR P 1807028
MANOJ G 1807030
SURYA P 1807052

In partial fulfillment for the award of the degree

of

BACHELOR OF TECHNOLOGY
in
INFORMATION TECHNOLOGY

COIMBATORE INSTITUTE OF TECHNOLOGY, COIMBATORE-641014

(Government Aided Autonomous Institution Affiliated to Anna University)

ANNA UNIVERSITY, CHENNAI 600025

 ABSTRACT
Chronic kidney disease (CKD) is a global health problem with high morbidity

and mortality rate, and it induces other diseases. Since there are no obvious symptoms

during the early stages of CKD, patients often fail to notice the disease. Early

detection of CKD enables patients to receive timely treatment to ameliorate the

progression of this disease. Machine learning models can effectively aid clinicians

achieve this goal due to their fast and accurate recognition performance. In this study,

we propose a machine learning methodology for diagnosing CKD. The CKD data set

was obtained from the University of California Irvine (UCI) machine learning

repository, which has a large number of missing values. KNN imputation was used to

fill in the missing values, which selects several complete samples with the most

similar measurements to process the missing data for each incomplete sample.

Missing values are usually seen in real-life medical situations because patients may

miss some measurements for various reasons. After effectively filling out the

incomplete data set, Five machine learning algorithms (logistic regression, random

forest, support vector machine, k-nearest neighbor, naive Bayes classifier) were used

to establish models.
 INTRODUCTION
CHRONIC kidney disease (CKD) is a global public health problem affecting
approximately 10% of the world’s population. The percentage of prevalence of CKD
in China is 10.8% , and the range of prevalence is 10%-15% in the United States.
According to another study, this percentage has reached 14.7% in the Mexican adult
general population. This disease is characterised by a slow deterioration in renal
function, which eventually causes a complete loss of renal function. CKD does not
show obvious symptoms in its early stages. Therefore, the disease may not be detected
until the kidney loses about 25% of its function . In addition, CKD has high morbidity
and mortality, with a global impact on the human body.

Machine learning refers to a computer program, which calculates and deduces

the information related to the task and obtains the characteristics of the corresponding
pattern . This technology can achieve accurate and economical diagnoses of diseases;
hence, it might be a promising method for diagnosing CKD.

We used KNN imputation to fill in the missing values in the data set, which
could be applied to the data set with the diagnostic categories are unknown.Logistic
regression (LOG), RF, SVM, KNN, naive Bayes classifier (NB) were used to establish
CKD diagnostic models on the complete CKD data sets. The models with better
performance were extracted for misjudgment analysis.

KNN imputation is used to fill in the missing values. To our knowledge, this is
the first time that KNN imputation has been used for the diagnosis of CKD. In
addition, building an integrated model is also a good way to improve the performance
of separate individual models. The proposed methodology might effectively deal with
the scene where patients are missing certain measurements before being diagnosed.
 LITERATURE SURVEY

PAPER 1]

AUTHOR AND JOURNAL :

GUOZHEN CHEN , CHENGUANG DING , YANG LI , XIAOJUN HU , XIAO LI , LI REN

IEEE

TITLE :

Prediction of Chronic Kidney Disease Using Adaptive Hybridized Deep

Convolutional Neural Network on the Internet of Medical Things Platform.

METHODOLOGY :

Adaptive hybridized Deep Convolutional Neural Network (AHDCNN)

CNN

Internet of medical things platform (IoMT)

DRAWBACKS :

MISSING VALUES ARE FILLED BY MEDIAN METHOD WHICH IS NOT PROMISING

PAPER 2]

AUTHOR AND JOURNAL :

ALVARO SOBRINHO, ANDRESSA C. M. DA S. QUEIROZ , MARIA ELIETE

PINHEIRO , AND ANGELO PERKUSICH

IEEE

TITLE :
Computer-Aided Diagnosis of Chronic Kidney Disease in Developing Countries:
A Comparative Analysis of Machine Learning Techniques

METHODOLOGY :

k-fold cross-validation method based on the Weka software

J48 DECISION TREE

DRAWBACKS :

ACCURACY IS ONLY ABOUT 95%

LIMITED VALUES IN DATASET

PAPER 3]

AUTHOR AND JOURNAL :

ERLEND HODNELAND,

EIRIK KEILEGAVLEN,

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING

TITLE :

Detection of CKD using Tissue deformation fields from dynamic MR imaging

METHODOLOGY :

Image registration method

USES DYNAMIC MR IMAGING

DRAWBACKS :

DATASET IS VERY SMALL

Paper 4]
AUTHOR AND JOURNAL :

AHMED AMIJOHARI,MOHD

HELMY Abd WAHAB

IEEE International Conference on Information Systems and Computer

Networks

TITLE :

Two Class Classification comparative experiments for CKD

METHODOLOGY :

Two class Decision Forest && Two class Neural Networks

DRAWBACKS :

NO OF SAMPLES IS LESS

STAGE OF THE DISEASE CAN’T BE POINTED

Paper 5]

AUTHOR AND JOURNAL :

BILAL KHAN , RASHID NASEEM , FAZAL MUHAMMAD , GHULAM ABBAS,

SUNGHWAN KIM

IEEE

TITLE :

An Empirical Evaluation of Machine Learning Techniques for Chronic Kidney

Disease Prophecy

METHODOLOGY :
Seven ML techniques together with NBTree, J48 are used.

(MAE), (RMSE), (RAE), (RRSE), recall

IMPROVEMENTS :

Severity of the disease can’t be predicted

PAPER 6]

AUTHOR AND JOURNAL :

Navaneeth Bhaskar and Suchetha M

IEEE

TITLE :

A Deep Learning-based System for Automated Sensing of Chronic Kidney

Disease

METHODOLOGY :

CNN-SVM integrated network

1-D Deep Learning convolution network

Used Saliva Samples

IMPROVEMENTS :

Only one Parameter is taken for prediction which decreases the prediction
capacity

PAPER 7]

AUTHOR AND JOURNAL :

Shubham Vashisth ,Ishika Dhall

International Conference on Cloud Computing

TITLE :

Chronic Kidney Disease (CKD) Diagnosis using Multi-

Layer Perceptron Classifier

METHODOLOGY :

Multi-Layer Perceptron Classifier

CMS data set is used

IMPROVEMENTS :

The dataset size can be increased

Accuracy can be increased

PAPAER 8]

AUTHOR AND JOURNAL :

N V Ganapathi Raju,

K Prasanna Lakshmi,

K. Gayathri Praharshitha

Chittampalli Likhitha

2019 ICICCS

TITLE :

Prediction of

chronic

kidney

disease (CKD)
using Data

Science

METHODOLOGY :

SVM, Random Forest,

XGBoost, Logistic

Regression, Neural

networks, Naive Bayes

Classifier.

IMPROVEMENTS :

The missing values can be filled using KNN.

Accuracy can be increased

PAPAER 9]

AUTHOR AND JOURNAL :

Yedilkhan Amirgaliyev , Shahriar Shamiluulu,

Azamat Serek

IEEE

TITLE :

Analysis of CKD using Machine Learning Techniques

METHODOLOGY :

Used Support Vector Machine(SVM)

IMPROVEMENTS :
Accuracy is only about 93%

An Integrated Model can be proposed

PAPER 10]

AUTHOR AND JOURNAL :

Gunarathne W.H.S.D, Perera K.D.M , Kahandawaarachchi K.A.D.C.P

IEEE International Conference on Bioinformatics and Bioengineering

TITLE :

Performance Evaluation on Machine Learning Classification Techniques and

Forecasting through Data Analytics for Chronic Kidney Disease (CKD)

METHODOLOGY :

Uses Multi cast Decision Tree Classifier

IMPROVEMENTS :

Accuracy is about 99.1%

They reduced the dataset size to 15 attributes

PAPER 11]

AUTHOR AND JOURNAL :

Ahmed J. Aljaaf, Dhiya Al-Jumeily, Hussein M. Haglan

IEEE Congress on Evolutionary Computation (CEC)

TITLE :
Early Prediction of Chronic Kidney Disease Using Machine Learning Supported
by Predictive Analytics

METHODOLOGY :

The Classification and Regression Tree.i.e.

RPART

Two Black box models SVM and MLP

IMPROVEMENTS :

No information about any kind of medications has been collected with this
data

PAPER 12]

AUTHOR AND JOURNAL :

Abdullah Al Imran, Md Nur Amin , Fatema Tuj Johora

International Conference on Innovation in Engineering and Technology (ICIET

TITLE :

Classification of Chronic Kidney Disease using Logistic Regression, Feedforward

Neural Network and Wide & Deep Learning

METHODOLOGY :

Logistic regression, feedforward neural networks and wide & deep learning to
diagnose CKD

IMPROVEMENTS :

They simply removed the datasets containing missing values(KNN can be used)

PAPER 13]

AUTHOR AND JOURNAL :

Hanyu Zhang, Che-Lun Hung, William Cheng-Chung Chu,Ping-Fang Chiu§ and
Chuan Yi Tang

IEEE International Conference on Bioinformatics and Biomedicine

TITLE :

Chronic Kidney Disease Survival Prediction with Artificial Neural Networks

METHODOLOGY :

Artificial Neural Network (ANN) models while applying to the survivability

prediction on Chronic Kidney Disease (CKD) patients.

IMPROVEMENTS :

They indicated that the dataset is much imbalanced(KNN can be used)

PAPER 14]

AUTHOR AND JOURNAL :

K.Shankar,P. Manickam, G. Devika,M. Ilayaraja

IEEE International conference on computational and computing research

TITLE :

Optimal Feature Selection for Chronic Kidney Disease Classification using Deep
Learning Classifier

METHODOLOGY :

Ant Lion Optimization (ALO) technique to choose optimal features for the
classification process.

Deep Neural Network (DNN)

IMPROVEMENTS :

Data mining procedures can be utilized as a part of training with enhancing

execution of classifiers, and the datasets are expanded