Scholarlink

Background <p>Diabetes and prediabetes significantly increase the risk of atherosclerotic cardiovascular disease (ASCVD), posing a major global health challenge. Although traditional ASCVD risk factors have been extensively studied, there is limited research on applying machine learning.</p> Methods <p> This study used data from the NHANES survey spanning 2007 to 2018, including 4,211 participants diagnosed with diabetes or prediabetes. Key variables were identified through univariate and multivariate logistic regression analyses. The dataset was randomly split into training and validation sets at a 7:3 ratio. Nine machine learning models (including CART, SVM, and GBM) were developed and evaluated using AUC, Brier scores, calibration curves, and decision curve analysis. Additionally, an online risk prediction platform was created to provide real-time ASCVD risk assessments, helping clinicians with early screening and intervention.</p> Results <p> Eight variables significantly associated with ASCVD risk were identified through univariate and multivariate logistic regression analyses, including age, waist circumference, poverty–income ratio, blood urea nitrogen, total cholesterol, systolic blood pressure, hypertension, and smoking status. Based on these predictors, the SVM model achieved AUC values of 0.831 in the training set and 0.859 in the validation set, demonstrating excellent discriminative ability. Calibration curves indicated good agreement between predicted and observed risks across different risk levels, while Brier scores further supported the overall predictive accuracy of the model. Decision curve analysis showed that the model provided substantial net clinical benefit. Collectively, these evaluation metrics highlight the strong predictive performance and practical utility of the SVM model.</p> Conclusion <p> The SVM model effectively predicts ASCVD risk in individuals with diabetes or prediabetes, emphasizing the importance of managing modifiable risk factors such as dyslipidemia, hypertension, smoking, and abdominal obesity. This model offers an effective tool for individualized risk assessment and early prevention, and the online platform further supports clinical application by enabling early identification and intervention for high-risk populations.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning-based prediction of atherosclerotic cardiovascular disease risk in adults with diabetes or prediabetes

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月4日

查看全文
  • Yang Li,
  • Bing Wang

摘要

Background

Diabetes and prediabetes significantly increase the risk of atherosclerotic cardiovascular disease (ASCVD), posing a major global health challenge. Although traditional ASCVD risk factors have been extensively studied, there is limited research on applying machine learning.

Methods

This study used data from the NHANES survey spanning 2007 to 2018, including 4,211 participants diagnosed with diabetes or prediabetes. Key variables were identified through univariate and multivariate logistic regression analyses. The dataset was randomly split into training and validation sets at a 7:3 ratio. Nine machine learning models (including CART, SVM, and GBM) were developed and evaluated using AUC, Brier scores, calibration curves, and decision curve analysis. Additionally, an online risk prediction platform was created to provide real-time ASCVD risk assessments, helping clinicians with early screening and intervention.

Results

Eight variables significantly associated with ASCVD risk were identified through univariate and multivariate logistic regression analyses, including age, waist circumference, poverty–income ratio, blood urea nitrogen, total cholesterol, systolic blood pressure, hypertension, and smoking status. Based on these predictors, the SVM model achieved AUC values of 0.831 in the training set and 0.859 in the validation set, demonstrating excellent discriminative ability. Calibration curves indicated good agreement between predicted and observed risks across different risk levels, while Brier scores further supported the overall predictive accuracy of the model. Decision curve analysis showed that the model provided substantial net clinical benefit. Collectively, these evaluation metrics highlight the strong predictive performance and practical utility of the SVM model.

Conclusion

The SVM model effectively predicts ASCVD risk in individuals with diabetes or prediabetes, emphasizing the importance of managing modifiable risk factors such as dyslipidemia, hypertension, smoking, and abdominal obesity. This model offers an effective tool for individualized risk assessment and early prevention, and the online platform further supports clinical application by enabling early identification and intervention for high-risk populations.