Predicting cardiovascular outcomes in Chinese patients with type 2 diabetes by combining risk factor trajectories and...

Area of Science:

• Cardiology
• Diabetology
• Machine Learning in Healthcare

Background:

• Cardiovascular complications pose a significant threat to Chinese patients with type 2 diabetes.
• Traditional risk models struggle with accurate prediction of these complications.
• There is a need for advanced methods to improve cardiovascular risk assessment in this population.

Purpose of the Study:

• To develop a dynamic prediction model for cardiovascular risk in Chinese patients with type 2 diabetes.
• To leverage machine learning and longitudinal data of risk factors for enhanced prediction accuracy.
• To compare the performance of the new model against existing risk prediction tools.

Main Methods:

• A cohort of 16,378 Chinese patients with type 2 diabetes was analyzed.
• A machine learning model (ML-CVD-C) was developed using baseline and four-year longitudinal data.
• The ML-CVD-C score's predictive performance was compared with ML-CVD-C (base), China-PAR, and PREVENT models.

Main Results:

• The ML-CVD-C score demonstrated superior discrimination (C-index 0.80) compared to other models (0.62-0.65).
• Significant improvements in risk reclassification were observed (NRI gains up to 57.7%, IDI gains up to 10.1%).
• Both dynamic trajectories and the machine learning algorithm were crucial for the enhanced performance.

Conclusions:

• The ML-CVD-C model, integrating dynamic risk trajectories and machine learning, substantially improves cardiovascular risk prediction accuracy in Chinese type 2 diabetes patients.
• This model offers a promising tool for personalized cardiovascular risk management.
• Dynamic risk factor analysis is key to advancing predictive capabilities in diabetes care.