January 22, 2026Open Access

Clinical Risk Prediction Models for Worsening Heart Failure Events and All-Cause Mortality in Adults with Mild-to-Moderate Chronic Kidney Disease

Resultado clave

EHR-based machine-learning models predicted worsening heart failure and all-cause mortality with AUCs of 0.887 and 0.875, respectively, in adults with mild-to-moderate CKD.

Puntos clave

This research aims to develop and validate models that predict worsening heart failure events and all-cause mortality in adults with mild-to-moderate chronic kidney disease.
Analyzed data from adults with mild-to-moderate chronic kidney disease receiving care from 2012 to 2021.
Outcomes were determined through electronic health records and validated algorithms.
Utilized machine learning (XGBoost) with an 80:20 train-test split and over 500 covariates derived from EHR.
Evaluated model performance using discrimination (AUC) and calibration across various subgroups.
The worsening heart failure model achieved an AUC of 0.887, indicating strong predictive ability.
The mortality model showed an AUC of 0.875, also reflecting robust prediction accuracy.
Model performance remained consistent across age, sex, and race/ethnicity, with slight decrease as CKD stage worsened.

PICO estructurado

Can EHR-based machine-learning models accurately predict worsening heart failure events and all-cause mortality in adults with mild-to-moderate CKD?

Población

375,495 adults with mild-to-moderate CKD (eGFR 30–59 ml/min/1.73m² or eGFR ≥60 with albuminuria), mean age 64±16 years, 54% women, 53% non-Hispanic White.

Intervención

EHR-based machine-learning models (extreme gradient boosting [XGBoost]) using >500 EHR-derived covariates

Resultado

Worsening heart failure (WHF) events (outpatient encounters, emergency department/observation stays, and hospitalizations) and all-cause mortalityhard clinical

EHR-based machine-learning models using >500 covariates accurately predict worsening heart failure and all-cause mortality in adults with mild-to-moderate CKD, supporting risk-stratified care.

Resultado numérico

Tasa de eventos absoluta: 0% vs 0%

Resumen

Abstract Aims To develop and internally validate electronic health record (EHR)–based machine-learning models to predict worsening heart failure (WHF) events across care settings and all-cause mortality among adults with mild-to-moderate chronic kidney disease (CKD). Methods and Results We studied adults with mild-to-moderate CKD (estimated glomerular filtration rate eGFR 30–59 ml/min/1.73m² or eGFR ≥60 with albuminuria) receiving care in a large health system from 2012–2021; outcomes were ascertained through 31 December 2022. Primary outcomes were (1) WHF events—outpatient encounters, emergency department (ED)/observation stays, and hospitalizations—identified using a validated natural language processing algorithm, and (2) all-cause mortality. Models (extreme gradient boosting XGBoost) used an 80:20 train–test split and 500 EHR-derived covariates. Discrimination (area under the curve AUC) and calibration (slope) were evaluated overall and across subgroups by age, sex, race and ethnicity, and CKD stage. Among 375,495 adults (mean age 64±16 years; 54% women; 53% non-Hispanic White; mean eGFR 76±26 ml/min/1.73m²), the WHF model achieved AUC 0.887 (95% CI 0.879–0.893) with calibration slope 0.955; the mortality model achieved AUC 0.875 (95% CI 0.868–0.883) with calibration slope 0.914 in the test set. Performance was consistent across age, sex, and race and ethnicity, with a slight decrement as CKD stage worsened. Conclusions EHR-based machine-learning models accurately predicted WHF and mortality in mild-to-moderate CKD with strong calibration across key subgroups. These models are positioned for EHR deployment to support risk-stratified cardiovascular-kidney-metabolic care—prioritizing guideline-directed therapies and care pathways for those at highest risk.

Leer artículo completoexternamente

Me gusta

Guardar

Ver artículo completo