A machine learning driven monogram for predicting chronic kidney disease stages 3-5.

Area of Science:

• Nephrology
• Medical Informatics
• Biostatistics

Background:

• Chronic kidney disease (CKD) is a major global health concern, contributing significantly to mortality.
• Accurate prediction models are crucial for early CKD detection and effective prevention strategies.
• Machine learning (ML) shows promise in advancing medical prediction capabilities.

Purpose of the Study:

• To introduce a novel ML-driven nomogram for early identification of individuals at risk for CKD stages 3-5.
• To develop and validate a predictive model using clinical and laboratory data.
• To assess the performance of ML models compared to traditional prediction methods.

Main Methods:

• Retrospective analysis of a large cohort of diagnosed CKD patients.
• Application of advanced ML algorithms, including feature selection and regression models (Linear Regression, Support Vector Machine).
• Development of an LR-based nomogram utilizing significant predictive factors (age, gender, medical history, lab results).

Main Results:

• 11.56% of 467 participants developed CKD stages 3-5 over a 9-year follow-up.
• Several factors demonstrated independent significant associations with CKD development (p < 0.05).
• The LR-based model achieved a high R-score of 0.954079, outperforming SVM and traditional models.

Conclusions:

• The ML-driven nomogram is a superior and valuable clinical tool for early CKD risk identification.
• This approach facilitates improved risk management and prevention of advanced CKD stages.
• Further research is needed to refine the model and validate its performance across diverse populations.