Comprehensive Summary
The research is primarily focused on the use of machine learning (ML) models to predict obstructive coronary artery disease (CAD) and to predict revascularization. An exploratory analysis was conducted on the Coronary Computed Tomographic Angiography for Selective Cardiac Catheterization (CONSERVE) study, which had branches that dealt with invasive coronary angiography (ICA), as well as coronary computed tomographic angiography (CCTA). The data from each branch was fed to machine learning models (based on extreme gradient boosting), in order to test the importance of different variables in predicting obstructive CAD and comparing results to the CAD consortium clinical score (CAD2), as well as in predicting revascularization. It was found that the ML models significantly outperformed the CAD2, and highlighted important risk factors as variables for predicting CAD, such as age, angina severity, and body mass index (BMI). The ML models also did fairly well in predicting revascularization, although not to the extent that they did for CAD.
Outcomes and Implications
Through the machine learning models utilized in this study, better models to predict obstructive CAD have been developed. This is critical to patients at risk for CAD, providing early warning methods, which provides patients and clinicians the ability to mitigate these conditions and any behaviors risking their health before the condition can progress. The study also highlights the critical role that imaging data plays in predicting serious conditions, as well as in identifying risk factors that can contribute to the development of diseases such as obstructive CAD. The research also concluded that both invasive coronary angiography (ICA) and coronary computed tomographic angiography (CCTA) are similar enough in results that either mode of imaging can be used for diagnostic imaging in cases of obstructive CAD and revascularization.