A Machine Learning-Based Approach for the Prediction of Acute Coronary Syndrome Requiring Revascularization

Area of Science:

• Cardiology
• Medical Informatics
• Machine Learning

Background:

• Early diagnosis of acute coronary syndrome (ACS) is crucial for timely intervention.
• Angina-like symptoms present a diagnostic challenge, requiring accurate differentiation between ACS and non-ACS patients.
• Machine learning offers potential for improving diagnostic accuracy in complex cardiovascular conditions.

Purpose of the Study:

• To develop and validate machine learning algorithms for predicting ACS requiring revascularization in patients with early angina-like symptoms.
• To identify key predictive features for ACS and assess the impact of missing data on model performance.
• To establish a reliable method for discriminating between ACS and non-ACS patients to guide treatment decisions.

Main Methods:

• Utilized data from 2344 ACS patients requiring revascularization and 3538 non-ACS patients.
• Applied standard algorithms, support vector machines, and linear discriminant analysis to 20 relevant ACS features.
• Investigated the impact of feature omission and data completeness on prediction accuracy using receiver operating characteristic (ROC) curves.

Main Results:

• Achieved an area under the curve (AUC) of 0.860 for predicting ACS requiring revascularization.
• Demonstrated that omitting informative but incomplete features and using more complete data improved prediction accuracy.
• Identified a reliable prediction method for 2.60% of non-ACS patients with 1.0 specificity, enabling safe medical treatment recommendations.

Conclusions:

• Machine learning algorithms can effectively predict acute coronary syndrome requiring revascularization in patients with early angina-like symptoms.
• Optimizing feature selection and handling missing data are critical for enhancing predictive model performance.
• The developed approach shows promise for safely discriminating non-ACS patients, allowing for appropriate medical management without compromising care for those needing revascularization.