NHANES‐Derived Machine Learning Model for Early Identification of Frailty Risk in CKM: Optimizing Resource Allocation Through Predictive Analytics

Background CKM encompasses a complex group of disease states involving cardiovascular, kidney, and metabolic dysfunction. Frailty is a significant health issue among older adults and is closely associated with adverse outcomes. However, there is currently a lack of predictive tools specifically designed for assessing frailty risk in middle‐aged and older patients with CKM Stages 0–3. Objective This study is aimed at developing a machine learning model to predict frailty risk in this population and to identify key predictive factors. Methods Using data from NHANES 2007–2018, a total of 6749 participants aged 40 years and older were included. Feature selection was conducted using LASSO regression, and the cohort was randomly divided into training and testing sets in a ratio of 7:3. Five machine learning models LR, LightGBM, XGBoost, RF, and DT were compared. Model performance was evaluated via 10‐fold cross‐validation, and the optimal model was selected based on accuracy, AUC, precision, recall, and F1 score. Feature importance was further explored using SHAP values. Results Among the 6749 participants, 1427 (21.14%) were identified as frail. LASSO regression identified 21 important variables, which were used to construct the five machine learning models. The RF model achieved the best performance in predicting frailty risk, as demonstrated by its AUC (0.90), accuracy (0.81), precision (0.54), recall (0.84), and F1 score (0.66). Decision curve analysis and calibration plots further confirmed the clinical utility of this approach. SHAP analysis revealed that the most significant predictors were PIR, antihypertensive medication use, hemoglobin, red blood cell count, albumin, and diabetes. Conclusion The machine learning model developed in this study can effectively predict the risk of frailty in middle‐aged and older adults with CKM Stages 0–3, providing a valuable tool for early identification of high‐risk individuals in clinical practice. This model may help optimize healthcare resource allocation, guide‐targeted interventions, and improve health management for patients with CKM.