A nine-variable prediction model (age, gender, mobility difficulty, chronic kidney disease, anemia, neutrophil, blood urea nitrogen, albumin, and C-reactive protein) built in 2893 patients with stage 4 CKM syndrome showed robust performance for predicting CVD mortality, and may enable clinicians to identify high-risk patients early and formulate individualized intervention strategies.
Nine risk factors were identified for predicting CVD mortality in stage 4 CKM syndrome patients.
The nomogram model showed comparable discrimination in training and validation sets.
The model performed best for predicting 15-year CVD mortality.
Calibration was comparable and decision curve analysis showed greater clinical benefit of the nomogram score than any single variable.
Participants with high nomogram scores had lower survival rates.
Random Forest and XGBoost performed best among six machine learning models used to validate the model.
SHAP analysis identified age and C-reactive protein as the largest contributors to CVD mortality prediction.
A nomogram and web-based calculator were developed as clinical decision support tools.
This study built a tool to estimate the risk of dying from cardiovascular disease in people with stage 4 cardiovascular-kidney-metabolic (CKM) syndrome, the most advanced stage of a condition in which heart, kidney, and metabolic problems occur together. Using data from 2893 patients, researchers selected nine factors: age, gender, difficulty with mobility, chronic kidney disease, anemia, neutrophil count, blood urea nitrogen, albumin, and C-reactive protein (a marker of inflammation). The resulting model separated higher-risk from lower-risk patients reasonably well, with similar performance in the group used to build it and the group used to test it (C-index 0.773 and 0.779). It was most accurate for predicting 15-year cardiovascular death (AUC 0.871). Patients with high scores had lower survival. Machine learning methods such as Random Forest and XGBoost performed well, and age and C-reactive protein were the most influential factors. The researchers also created a nomogram and a web-based calculator. This research suggests that routinely available clinical and blood test information could help identify patients with advanced CKM syndrome who are at especially high risk of cardiovascular death, which might support earlier and more individualized care planning. The tool was developed and tested within the same dataset through random splitting, so further testing in other populations would be needed to confirm how well it works elsewhere.
Zhang P, Wang S, Yang H, He Y, Xu Z, Wu T, et al.. (2026). Machine learning based predictive model for cardiovascular disease mortality risk in patients with stage 4 cardiovascular-kidney-metabolic syndrome.. Medicine. https://doi.org/10.1097/MD.0000000000051266