An evolutionary machine learning algorithm for cardiovascular disease risk prediction.

Area of Science:

• Cardiology
• Biostatistics
• Machine Learning

Background:

• Cardiovascular disease (CVD) poses a significant global health challenge.
• Accurate risk assessment is crucial for effective CVD prevention strategies.

Purpose of the Study:

• To develop a novel, accurate, and interpretable risk assessment model for predicting cardiovascular disease (CVD) events.
• To leverage a Genetic Algorithm (GA) for enhanced CVD risk prediction using the Isfahan Cohort Study (ICS) database.

Main Methods:

• A population-based prospective cohort study (ICS) involving 6,504 Iranian adults (≥35 years) followed for 10 years (2001-2010).
• Development of a CVD risk score using a Genetic Algorithm (GA).
• Comparison of the GA-based model with established statistical and machine learning (ML) methods, including Deep Learning (CNN).

Main Results:

• The GA-based eXplanaible Persian Atherosclerotic CVD Risk Stratification (XPARS) model demonstrated superior performance.
• XPARS with eight features achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.76, outperforming traditional models (e.g., Framingham AUROC=0.633, PROCAM AUROC=0.683).
• A Deep Learning (CNN) model achieved an AUROC of 0.74, while XPARS with four features had an AUROC of 0.72.

Conclusions:

• Genetic Algorithm-driven risk models substantially enhance CVD prediction compared to conventional approaches.
• The developed GA model (XPARS) is clear, interpretable, and suitable as a replacement for traditional statistical methods in CVD risk assessment.