Retinal Microvascular Characteristics-Novel Risk Stratification in Cardiovascular Diseases

Area of Science:

• Ophthalmology
• Cardiology
• Biomedical Engineering

Background:

• Cardiovascular diseases (CVDs) cause significant mortality in the EU.
• Retinal microvasculature shows promise as a biomarker for cardiovascular risk.
• Existing CVD risk scores have variable effectiveness.

Purpose of the Study:

• Identify retinal microvascular features linked to coronary heart diseases (CHDs).
• Assess the utility of these features in a CHD screening algorithm with traditional risk factors.

Main Methods:

• Cross-sectional study with 120 participants (36 CHD patients, 84 controls).
• Utilized optical coherence tomography angiography (OCTA) to analyze retinal microvasculature.
• Compared logistic regression with machine learning (k-NN, Naive Bayes, SVM) for CHD classification.

Main Results:

• Machine learning models outperformed conventional logistic regression.
• Support Vector Machine (SVM) achieved the highest accuracy at 86% for CHD identification.
• Accuracies ranged from 78.7% (logistic regression) to 86% (SVM).

Conclusions:

• Machine learning algorithms combined with retinal microvascular data enhance CHD identification.
• Novel risk scores integrating OCTA parameters could improve CHD screening.
• Retinal imaging offers a valuable adjunct to traditional cardiovascular risk assessment.