Predicting 3-year all-cause mortality in patients undergoing hemodialysis using machine learning

Area of Science:

• Nephrology
• Cardiology
• Medical Informatics

Background:

• Hemodialysis (HD) patients have low survival rates, with cardiovascular events being the primary cause of mortality.
• Existing risk models often overlook electrocardiogram (ECG) findings, a crucial aspect of cardiovascular health.

Purpose of the Study:

• To develop and validate a novel risk prediction model for all-cause mortality in patients undergoing hemodialysis.
• To incorporate electrocardiogram (ECG) parameters into a risk model for enhanced predictive accuracy.

Main Methods:

• A cohort of 454 patients undergoing HD was analyzed from April 2008 to March 2021.
• Multivariate Cox regression identified independent predictors of mortality, including age, serum albumin, stroke history, atrial fibrillation, and corrected QT interval.
• A nomogram-based risk model was developed and validated using area under the curve (AUC) and calibration plots.

Main Results:

• The 3-year follow-up revealed a 21.5% mortality rate (98 deaths).
• The novel risk model demonstrated good predictive performance with an AUC of 0.83 (95% CI, 0.79-0.87), 80.1% sensitivity, and 75.6% specificity.
• Cross-validation confirmed the model's robustness with an AUC of 0.82.

Conclusions:

• The developed risk model effectively stratifies hemodialysis patients based on their 3-year all-cause mortality risk.
• This tool can aid in early identification of high-risk patients, enabling personalized treatment and safer prescriptions.