Enhancing machine learning-based forecasting of chronic renal disease with explainable AI.

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Predictive Analytics

Background:

• Chronic Renal Disease (CRD) poses a significant healthcare challenge, necessitating early and accurate prediction for effective treatment.
• Current diagnostic methods may lack the speed and accuracy required for timely intervention.

Purpose of the Study:

• To develop and evaluate an end-to-end predictive model for the binary classification of CRD.
• To enhance model performance through hyperparameter optimization and compare various machine learning algorithms.
• To integrate Explainable Artificial Intelligence (XAI) for improved model interpretability.

Main Methods:

• Implementation of an end-to-end machine learning pipeline for CRD prediction.
• Utilized GridSearchCV for hyperparameter optimization.
• Evaluated Random Forest, Extra Trees Classifier, Logistic Regression (L2 penalty), and Artificial Neural Networks (ANN).
• Applied XAI techniques like LIME and SHAP for model interpretability.

Main Results:

• Achieved a high predictive accuracy of 99.07% across multiple models.
• Logistic Regression with L2 penalty demonstrated superior and consistent performance.
• XAI techniques provided insights into the decision-making processes of the predictive models.

Conclusions:

• The developed end-to-end predictive model shows significant potential for early CRD detection.
• Optimized machine learning models, particularly logistic regression, offer high accuracy in CRD classification.
• Integration of XAI enhances trust and understanding of predictive models in clinical settings, supporting informed decision-making.