Development and Validation of a Machine Learning Model for Incident Heart Failure Prediction in Chronic Kidney Disease: A Multicenter Cohort Study - PubMed
3 hours ago
- #machine-learning
- #chronic-kidney-disease
- #heart-failure
- Study developed and validated a machine learning model to predict incident heart failure (HF) in chronic kidney disease (CKD) patients.
- Used data from 52,251 CKD patients in China for development and 21,798 Chinese patients plus 3,323 UK Biobank participants for external validation.
- Extreme gradient boosting model outperformed others with AUC of 0.879 in Chinese cohort and 0.851 in UK Biobank cohort.
- Simplified model includes 9 predictors, with estimated glomerular filtration rate and albuminuria ranked as most important features.
- Model showed significant improvement over existing risk scores (Atherosclerosis Risk in Communities and Predicting Risk of Cardiovascular Disease Events Equation).
- Web-based calculator deployed for clinical use: https://clinician.shinyapps.io/HF_Risk_Predictor_for_CKD/.
- Conclusions suggest the model may help predict HF risk in CKD patients, though further validation across diverse populations is needed.