Optimized ensemble model for accurate prediction of cardiac vascular calcification in diabetic patients

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiology

Background:

• Cardiovascular diseases (CVD) pose a significant risk to diabetic patients.
• Cardiac vascular calcification (CVC) is a critical predictor of CVD in diabetes.
• Current machine learning (ML) and artificial intelligence (AI) methods for CVC prediction face limitations including small sample sizes and high computational costs.

Purpose of the Study:

• To develop an advanced ML model for improved prediction of CVC in diabetic individuals.
• To address the challenges of data limitations and computational complexity in existing AI methods for CVC detection.

Main Methods:

• A novel Simple linear iterative clustering-based Ensemble Artificial Neural Network (SLIC-EANN) model was developed.
• The model utilizes biochemical, imaging (from Coronary computed tomography angiography - CCTA), and clinical data.
• Preprocessing included image normalization, resizing, and augmentation; calcification localization used the Simple Linear Iterative Clustering (SLIC) algorithm; and classification employed an Ensemble Artificial Neural Network (EANN) integrating Support Vector Machine (SVM), Gradient Boosting (GB), and Decision Tree (DT).

Main Results:

• The SLIC-EANN model achieved a high prediction accuracy of 98.7%.
• The model demonstrated a low error rate of 1.3%.
• Performance significantly outperformed existing CVC prediction techniques.

Conclusions:

• The proposed SLIC-EANN model offers superior prediction performance for CVC in diabetic patients.
• This AI-driven approach shows promise for enhancing cardiovascular risk assessment in diabetes management.