Machine Learning Based Risk Prediction for Major Adverse Cardiovascular Events

Area of Science:

• Cardiovascular medicine
• Medical informatics
• Machine learning in healthcare

Background:

• Major adverse cardiovascular events (MACE), including myocardial infarction and stroke, lead to significant hospitalizations and mortality.
• Early identification of at-risk patients is crucial for implementing preventive interventions.
• Developing robust risk prediction tools is essential for improving patient outcomes.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting the 5-year risk of MACE.
• To leverage electronic medical record data for comprehensive cardiovascular risk assessment.
• To identify key features for accurate MACE prediction.

Main Methods:

• Utilized electronic medical records from over 128,000 patients, with 29,262 diagnosed with MACE.
• Applied feature selection techniques (filter and embedded methods) to identify 826 relevant features.
• Trained and evaluated various machine learning models on the prepared dataset.

Main Results:

• A random forest model demonstrated superior calibration and discriminative ability.
• The best performing model achieved an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.88 on a test dataset.
• The models showed excellent predictive performance in the evaluated test data.

Conclusions:

• The developed machine learning models exhibit excellent performance for 5-year MACE risk prediction.
• Further prospective studies are required to validate the clinical utility and benefit of these models.
• These models hold potential for early detection and prevention of cardiovascular events.