Machine Learning Hybrid Model for the Prediction of Chronic Kidney Disease.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Nephrology

Background:

• Chronic kidney disease (CKD) is a leading cause of death globally, necessitating advanced diagnostic tools.
• Current diagnostic methods, while effective, can be augmented by the high accuracy of machine learning (ML) approaches.
• The increasing prevalence of CKD underscores the need for efficient and accurate predictive models.

Purpose of the Study:

• To develop and evaluate a hybrid machine learning model for accurate prediction of chronic kidney disease (CKD).
• To compare the performance of the proposed hybrid model against established ML classifiers using the UCI chronic kidney disease dataset.
• To identify the most effective ML classification techniques for CKD diagnosis, addressing challenges like overfitting.

Main Methods:

• A hybrid ML model was constructed using Gaussian Naïve Bayes, gradient boosting, and decision tree as base classifiers, with random forest as the meta-classifier.
• Pearson correlation was employed for feature selection within the proposed model.
• The model was trained and evaluated on the UCI chronic kidney disease dataset, comparing its accuracy against individual classifiers.

Main Results:

• The proposed hybrid model achieved a perfect accuracy of 100% in predicting CKD.
• Individual models showed strong performance: gradient boosting (99%), random forest (98%), and decision tree (96%).
• The hybrid approach demonstrated superior predictive capability and effectively mitigated overfitting.

Conclusions:

• The developed hybrid machine learning model offers a highly accurate and effective solution for diagnosing chronic kidney disease (CKD).
• This approach represents a significant advancement in leveraging ML for early and precise CKD detection.
• The study highlights the potential of hybrid ML models to overcome limitations of single algorithms in complex medical diagnoses.