Predicting Hypoproteinemia Among Patients Undergoing Maintenance Hemodialysis: A Development and Validation Study...

Area of Science:

• Nephrology
• Medical Informatics
• Biostatistics

Background:

• Maintenance hemodialysis is a critical treatment for end-stage renal disease (ESRD).
• Hypoproteinemia is a common complication in ESRD patients undergoing hemodialysis.
• Accurate prediction of hypoproteinemia is essential for timely clinical intervention.

Purpose of the Study:

• To develop and evaluate machine learning-based prediction models for hypoproteinemia in maintenance hemodialysis patients.
• To identify independent risk factors associated with hypoproteinemia in this patient population.
• To assess the performance of different machine learning algorithms in predicting hypoproteinemia risk.

Main Methods:

• A cohort of 468 maintenance hemodialysis patients was analyzed.
• Univariate analysis was employed to identify independent risk factors for hypoproteinemia.
• Machine learning models including Random Forest (RF), Support Vector Machine, and Logistic Regression (LR) were trained and validated using k-fold cross-validation and grid search.

Main Results:

• The overall incidence of hypoproteinemia was 30.8%.
• Significant differences between hypoproteinemia and non-hypoproteinemia groups were observed in 18 aspects, including age, weight, dialysis duration, and frequency.
• The RF model achieved the highest prediction accuracy (0.924), outperforming other models.

Conclusions:

• Machine learning models, particularly Random Forest, can effectively predict hypoproteinemia risk in ESRD patients on maintenance hemodialysis.
• The developed prediction model can assist healthcare professionals in identifying at-risk patients.
• Early identification facilitates improved screening, primary prevention, and timely intervention strategies for hypoproteinemia in ESRD patients.