Using Data Mining Techniques to Predict Chronic Kidney Disease: A Review Study.

Area of Science:

• Nephrology
• Data Science
• Medical Informatics

Background:

• Chronic kidney disease (CKD) is a significant and escalating global health concern.
• Early diagnosis, effective management, and control are crucial for mitigating CKD's impact.
• Data mining and classification methods offer promising avenues for predicting kidney disease progression.

Purpose of the Study:

• To systematically review and analyze classification methods used in predicting kidney disease.
• To identify the most frequently employed data mining techniques and significant risk factors in CKD prediction studies.
• To evaluate the performance of different data mining models in the context of kidney disease prediction.

Main Methods:

• A comprehensive literature search was conducted for articles published in English between 2016 and 2021.
• Studies focusing on classification methods for kidney disease prediction were included.
• Frequency analysis of data mining techniques and identified risk factors was performed.

Main Results:

• Support Vector Machine, Naive Bayes, and K-Nearest Neighbor were the most frequent data mining techniques identified.
• Random Forest, Neural Network, and Decision Tree were also commonly utilized.
• Key risk factors frequently associated with CKD included albumin, age, red blood cells, pus cells, and serum creatinine.
• The Random Forest technique demonstrated the highest predictive performance (yields).

Conclusions:

• Data mining models, particularly Random Forest, are effective tools for predicting chronic kidney disease.
• Identifying key risk factors like albumin and age is vital for early detection and intervention strategies.
• Further research utilizing larger datasets is recommended for enhanced analysis and validation of CKD risk factors.