Noncommunicable diseases (NCDs), especially cardiovascular diseases (CVDs) and diabetes, are leading causes of death worldwide.

Our recent publication “Interpretable data-driven approach based on feature selection methods and GAN-based models for cardiovascular risk prediction in diabetic patients” is aimed to develop an interpretable machine learning (ML) model to predict 10-year CVD risk in older individuals with type 1 diabetes (T1D) using data from the Steno Diabetes Center Copenhagen. Various ML models, including KNN, decision tree, random forest, and multilayer perceptron (MLP), were used. Synthetic data generated with CTGAN improved model performance, and feature selection techniques identified key risk factors.

The MLP model performed best, with a mean absolute error of 0.0088. Key risk factors identified were age, HbA1c, and albuminuria.

This study helps in early intervention and treatment to prevent CVDs in T1D patients.

Discover more download the full issue here.