An efficient ensemble based machine learning approach for predicting Chronic Kidney Disease.

Area of Science:

• * Application of machine learning in medical diagnostics.
• * Development of predictive models for chronic diseases.

Background:

• * Chronic kidney disease (CKD) poses a significant health risk, potentially leading to kidney failure.
• * Early detection of CKD is crucial for effective treatment and management.
• * Machine learning (ML) demonstrates high reliability in early disease diagnosis.

Purpose of the Study:

• * To predict the onset of CKD utilizing various ML classification algorithms.
• * To evaluate the efficacy of ML models on a publicly available CKD dataset from UCI ML Repository.

Main Methods:

• * Twelve ML classification algorithms were employed with the full feature set.
• * The Synthetic Minority Over-Sampling technique (SMOTE) was applied to address class imbalance.
• * K-fold cross-validation was used to assess model performance, comparing results with and without SMOTE.

Main Results:

• * An ensemble stacking classifier achieved a 99.5% accuracy rate.
• * The study identified Support Vector Machine, Random Forest, and Adaptive Boosting as top-performing classifiers.

Conclusions:

• * An ensemble learning approach, combining top ML classifiers after SMOTE balancing, significantly improved CKD prediction.
• * This validated technique offers a promising, less intrusive, and cost-effective method for disease detection.
• * The proposed methodology has potential applications for early diagnosis across various medical conditions.