Meta-analysis evaluates machine learning models against TIMI and GRACE scores for predicting cardiovascular events, indicating ML may enhance patient outcomes.
Key Points
Machine learning models significantly outperform traditional TIMI and GRACE scores in predicting acute coronary syndrome.
In particular, Random Forest and XGBoost showed AUC values of 0.99 and 0.98, respectively, highlighting their predictive power.
This meta-analysis followed PRISMA guidelines, analyzing data from 50 studies involving over 1.5 million patients across various cohorts.
Adopting explainable AI and enhancing clinician training are recommended to integrate machine learning into electronic health records effectively.