Accepted for/Published in: Journal of Medical Internet Research
Date Submitted: Nov 16, 2024
Date Accepted: Apr 28, 2025
The Machine Learning Models in Major Cardiovascular Adverse Events Prediction Based on Coronary Computed Tomography Angiography: A Systematic Review
ABSTRACT
Background:
Coronary computed tomography angiography (CCTA) has emerged as the first-line non-invasive imaging test for patients at high risk of coronary artery disease (CAD), combined with machine learning (ML), it provides more valid evidence for the diagnosis of major adverse cardiovascular events (MACE) based on radiomic features.
Objective:
To evaluate the diagnostic value of ML models constructed using radiomic features extracted from coronary CT angiography (CCTA) in predicting MACE.
Methods:
We conducted a comprehensive search across five online databases for studies that utilized ML models to predict MACE via CCTA, exploring the correlation between radiomic features and MACE endpoints. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was employed to assess the risk of bias in the ML models, while the Radiomics Quality Score (RQS) was used to evaluate the methodological quality of the radiomics prediction model development and validation. Meta-analysis was performed using Meta-DiSc software (version 1.4), which included the I² score and Cochran’s Q test, along with StataMP 17 to assess heterogeneity and publication bias. Due to the high heterogeneity observed, subgroup analyses were conducted based on different model groups.
Results:
A total of 10 studies were included in the analysis. The pooled area under the receiver operating characteristic curve (AUROC) for ML models predicting MACE was 0.7879 in the training set and 0.7981 in the testing set. Logistic regression (LR), the most commonly used algorithm, achieved an AUROC of 0.8229 in the testing group and 0.7983 in the training group.
Conclusions:
The performance of ML models for predicting MACE was found to be superior to that of general models based on basic feature extraction and integration from CCTA. Specifically, LR-based ML diagnostic models show significant clinical potential, especially when combined with clinical features and worths further validation through more clinical trials. Clinical Trial: CRD42024596364
Citation
Per the author's request the PDF is not available.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.