Sign in Subscribe
Type 2 diabetes

Predictive Model for Diabetic Complications Identifies Key Risk Factors

Strategic Insights

— 1 min read

Researchers from Zhejiang Chinese Medical University have developed a predictive model using machine learning algorithms to identify patients at high risk of developing diabetic complications, specifically peripheral vascular disease and diabetic foot. The study analyzed data from 1240 patients with type 2 diabetes and its complications, identifying 17 key indicators that contribute to the development of these conditions.

Key Takeaways:

  • The predictive model, developed using a comprehensive weighting approach, was able to accurately identify high-risk patients for diabetic complications, achieving an AUC value of 0.79 and an accuracy rate of 0.78 for peripheral vascular disease.
  • The model also identified fibrinogen and C-reactive protein as statistically significant risk factors for diabetic complications.
  • Feature importance analysis using SHAP algorithm highlighted the significant influence of these features in identifying diabetic complications.
  • The study concludes that the predictive model can be employed as a tool to assist in reducing the incidence of diabetic complications.
  • Key indicators identified through the comprehensive weighting approach include fibrinogen, C-reactive protein, and cardiac structural parameters.
  • The study demonstrates the potential of machine learning algorithms in predicting and identifying risk factors for diabetic complications.
  • Researchers from Zhejiang Chinese Medical University, including Guangrong Tao, Yifeng Pan, Bing Chen, Chao Zheng, and Gehong Li, contributed to the study.

Statistics:

  • 1240 patients with type 2 diabetes and its complications were analyzed in the study.
  • The predictive model achieved an AUC value of 0.79 for peripheral vascular disease and 0.89 for diabetic foot.
  • The model's accuracy rate was 0.78 for peripheral vascular disease and 0.80 for diabetic foot.
  • Fibrinogen and C-reactive protein showed statistically significant differences between groups, with a SHAP feature importance analysis highlighting their significant influence in identifying diabetic complications.
  • The study used 27 indicators to rank the importance of features contributing to diabetic complications.

Sources:

  • The predictive model and risk factor identification for peripheral vascular disease and diabetic foot in diabetes based on machine learning models and explainable algorithms. Medicine, 2025;104(40).
  • Lippincott Williams & Wilkins, Two Commerce Sq, 2001 Market St, Philadelphia, PA 19103, USA. (Elsevier - www.elsevier.com; Medicine - www.journals.elsevier.com/medicine/)
  • Cardiovascular Week. October 20, 2025; p 81.
  • NewsRx. Data from Zhejiang Chinese Medical University Advance Knowledge in Vascular Diseases and Conditions (The predictive model and risk factor identification for peripheral vascular disease and diabetic foot in diabetes based on machine learning ...).

Read more