Reviewing the use and quality of machine learning in developing clinical prediction models for cardiovascular disease

Area of Science:

• Cardiology
• Medical Informatics
• Artificial Intelligence in Healthcare

Background:

• Cardiovascular disease (CVD) is a leading global cause of mortality, encompassing conditions like heart attacks and strokes.
• Early identification of high-risk individuals for CVD enables timely intervention, often with statin therapy, improving patient outcomes.
• Current clinical prediction models (CPMs) rely on statistical analysis of risk factors like BMI and family history.

Purpose of the Study:

• To review and compare existing clinical prediction models (CPMs) for cardiovascular disease (CVD) risk assessment.
• To analyze the emerging role and performance of machine learning (ML) approaches in predicting CVD risk.
• To evaluate the potential of ML to outperform traditional statistical CPMs in identifying individuals at high risk for CVD.

Main Methods:

• Review of current literature on clinical prediction models (CPMs) for cardiovascular disease (CVD).
• Analysis of studies comparing traditional statistical CPMs with machine learning (ML) based models.
• Evaluation of the performance metrics and methodologies of both CPMs and ML approaches in CVD risk prediction.

Main Results:

• Machine learning (ML) based approaches consistently demonstrate superior performance compared to the newest non-ML clinical prediction models (CPMs).
• While current non-ML CPMs are effective, ML models offer enhanced accuracy in identifying individuals at high risk for cardiovascular disease (CVD).
• The review highlights the significant potential of ML to advance CVD risk prediction and early intervention strategies.

Conclusions:

• Machine learning (ML) models show significant potential to outperform traditional clinical prediction models (CPMs) in cardiovascular disease (CVD) risk assessment.
• Despite promising results, further research and validation are required before ML can be recommended for widespread clinical implementation over existing CPMs.
• The findings underscore the need for continued investigation into ML applications for improving early detection and management of cardiovascular disease.