Machine learning to predict cardiovascular risk.

Area of Science:

• Cardiovascular disease epidemiology
• Machine learning in healthcare
• Predictive analytics in medicine

Background:

• Cardiovascular diseases remain a leading cause of mortality globally.
• Accurate cardiovascular risk assessment is crucial for timely intervention and prevention strategies.
• Existing risk scales like SCORE and REGICOR have limitations in predictive accuracy.

Purpose of the Study:

• To evaluate the predictive performance of 15 distinct machine learning (ML) methods for cardiovascular risk estimation.
• To compare the efficacy of these ML methods against established risk stratification scales (SCORE and REGICOR).
• To identify superior ML algorithms for enhancing cardiovascular event prediction in a large patient cohort.

Main Methods:

• A cohort of 38,527 patients from the Spanish ESCARVAL RISK study with a 5-year follow-up was analyzed.
• Cardiovascular risk was calculated using 15 ML algorithms and the SCORE and REGICOR scales.
• Performance metrics included Area Under the Receiver Operating Curve (AUC), C-index, diagnostic accuracy, sensitivity, and specificity.

Main Results:

• Quadratic discriminant analysis achieved the highest predictive capacity (AUC = 0.7086), followed by Naive Bayes and neural networks.
• Traditional scales SCORE and REGICOR demonstrated significantly lower predictive capacity (AUC = 0.63), ranking 11th and 12th.
• Seven ML methods exhibited a 7% higher AUC, alongside improved sensitivity and specificity, compared to SCORE and REGICOR.

Conclusions:

• Ten of the 15 evaluated machine learning methods significantly outperform the commonly used SCORE and REGICOR scales in predicting cardiovascular events.
• ML-based approaches offer superior classification indicators for cardiovascular risk assessment.
• The findings strongly support the integration of machine learning into the development of next-generation cardiovascular risk prediction tools.