Objective:To construct a frailty risk prediction model for elderly patients with CHD based on machine learning, to address the limitations of existing tools and provide evidence-based support for clinical practice.Methods:A retrospective study was conducted on elderly CHD patients hospitalized at the Heart Center of the First Affiliated Hospital of Henan University of Chinese Medicine from September 2023 to March 2024. Meta-analysis and expert meetings were used to identify the risk factors for frailty in elderly CHD patients. Three machine learning algorithms, Logistic Regression, Random Forest, and Support Vector Machine, were used to construct predictive models using R 4.3.1 software. The predictive performance of the models was evaluated using sensitivity, specificity, accuracy, and the area under the receiver operating characteristic curve (AUC).Results:A total of 490 elderly CHD patients were included, with 267 males and 223 females, and an age of (71.02 ± 7.73) years. Among them, 160 patients (32.7%) developed frailty. Of the constructed models, the Random Forest model demonstrated the best predictive performance, with an accuracy of 0.703, recall of 0.629, and F1 score of 0.741, while the AUC was 0.811 (95% CI 0.762-0.850). Conclusions:The Random Forest model exhibited good predictive performance in assessing frailty risk in elderly CHD patients, with high accuracy and reliability. Future external validation studies can further assess its applicability and stability in different populations.

Objective:To construct a frailty risk prediction model for elderly patients with CHD based on machine learning, to address the limitations of existing tools and provide evidence-based support for clinical practice.Methods:A retrospective study was conducted on elderly CHD patients hospitalized at the Heart Center of the First Affiliated Hospital of Henan University of Chinese Medicine from September 2023 to March 2024. Meta-analysis and expert meetings were used to identify the risk factors for frailty in elderly CHD patients. Three machine learning algorithms, Logistic Regression, Random Forest, and Support Vector Machine, were used to construct predictive models using R 4.3.1 software. The predictive performance of the models was evaluated using sensitivity, specificity, accuracy, and the area under the receiver operating characteristic curve (AUC).Results:A total of 490 elderly CHD patients were included, with 267 males and 223 females, and an age of (71.02 ± 7.73) years. Among them, 160 patients (32.7%) developed frailty. Of the constructed models, the Random Forest model demonstrated the best predictive performance, with an accuracy of 0.703, recall of 0.629, and F1 score of 0.741, while the AUC was 0.811 (95% CI 0.762-0.850). Conclusions:The Random Forest model exhibited good predictive performance in assessing frailty risk in elderly CHD patients, with high accuracy and reliability. Future external validation studies can further assess its applicability and stability in different populations.