Enhancing Accuracy on Chronic Kidney Disease Detection - Machine Learning with Resampling and Missing Value Treatment

Aug 6, 2026

Upload by Patrick

Chronic kidney disease requires early identification to enable timely treatment and prevention. This study improves prediction performance by handling missing data through linear interpolation and addressing class imbalance via SMOTE resampling. Feature selection combines Pearson correlation and Principal Component Analysis to reduce noise and improve model separability. Classification is performed using Support Vector Machine and Logistic Regression on a UCI Machine Learning dataset for chronic kidney disease, with multiple benchmark test scenarios. The best averaged results use SMOTE plus PCA, achieving 98.8% accuracy, 100% precision, and 98.77% F1-score.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Problem background and diagnosis challenges

Data science and predictive analysis for healthcare

Classification approach for kidney disease

Enhancing accuracy on chronic-kidney disease detection using machine learning with technique of resampling and missing value treatment Muhammad Raihan Wibowo1, Irma Palupi*1 School of Computing, Telkom University , Indonesia1 Article Info Keywords: Chronic Kidney Disease, Support Vector Machine, Logistic Regression, Principal Component Analysis, SMOTE Article history: Received: June 23 , 2023 Accepted: September 04, 2023 Published: November 30 , 2023 Cite: M. R. Wibowo and I. Palupi, “Enhancing Accuracy on Chronic-Kidney Disease Detection Using Machine Learning with Technique of Resampling and Missing Value Treatment”, KINETIK, vol. 8, no. 4`, Nov. 2023. [https://doi.org/10.22219/kinetik.v8i4.1761](https://doi.org/10.22219/kinetik.v8i4.1761) *Corresponding author. Irma Palupi E-mail address: [irmapalupi@telkomuniversity.ac.id](irmapalupi@telkomuniversity.ac.id) Abstract Chronic kidney disease is one of the deadliest diseases in the world. It is important to identify chronic kidney disease at an early stage, so that treatment and prevention can be carried out early. This study used linear interpolation method to treat the missing values, resampling using SMOTE method, and several feature selection methods, such as Pearson’s correlation coefficient and Principal component analysis. For the classification methods, Support Vector Machine and Logistic Regression were used to build prediction models for chronic kidney disease based on dataset on UCI Machine Learning. To measure the performance of the model, several test scenarios were tested out so it can be compared to the previous research on the detection of chronic kidney disease, which is used as a benchmark for this study. The best result from the experiment is obtained from the scenario of resampling using SMOTE and feature selection using Principal Component Analysis with averaged accuracy, precision, and f1-score respectively are 98,8%, 100%, dan 98,77% . 1. Introduction Chronic Kidney Disease (CKD) is a type of kidney disease that causes gradual decline in kidney function. This phenomenon can be observed over months or even years due to varying patient lifestyles. CKD is also known as renal failure, and according to current medical statistics, 10% of the world’s population suffers from chronic kidney disease. In 2005, approximately 58 million people died, and according to the World Health Organization (WHO), 35 million of those deaths were related to chronic disease [1] . The diagnosis of CKD typically begins with clinical data, laboratory tests, imaging studies, and ultimately, biopsy. Although biopsy is a standard diagnostic test, it has several drawbacks, such as being invasive, expensive, time-consuming, and occasionally risky. For instance, if a biopsy is performed, patients may experience facial swelling, fear of surgery, and potential misdiagnosis. Imaging techniques (such as mammography, sonography, and renal MRI) have been used to detect this disease for many years. However, there are limitations to their use, including concerns about radiation exposure. Despite these risks, the information obtained from imaging is not sufficient for diagnosing CKD [1] . A data science solution for the analysis of healthcare data is prospective to help saving lives and improve the quality of life. Data is one of the most crucial resources for researchers in the fields of health sciences and medicine. Through those sample data , analysts search for trends, patterns, and similarities to invent new treatments or strengthen existing ones. Providing sample data may be expensive , and rare strains (or fewer samples) may not appear to have enough information to make the prediction statistically significant. Thus making the learning process from the dataset more challenging to use appropriately [2] . Predictive analysis is a method that utilizes various techniques , such as machine learning, data mining, and statistics to forecast future events. In th

FAQs

How does the study handle missing values in the dataset?

It applies a linear interpolation method to fill in missing values before modeling.

What resampling technique is used to improve learning performance?

The study uses SMOTE (Synthetic Minority Over-sampling Technique) to resample the training data.

Which feature selection and modeling scenario achieved the best results?

The best performance comes from using SMOTE for resampling combined with Principal Component Analysis for feature selection, tested with averaged accuracy, precision, and F1-score.

Aug 6, 2026

Upload by Patrick