Why the study?
Major adverse cardiac events significantly impact perioperative morbidity and mortality, creating a need to optimize risk prediction.
Does a fully automated multimodal deep learning system integrating CCTA and clinical data improve 30-day MACE prediction compared to RCRI and CAD-RADS alone in patients undergoing elective non-cardiac surgery?
Population
639 patients undergoing CCTA for perioperative risk assessment before elective non-cardiac surgery
Comparison
Multimodal DL system vs RCRI and CAD-RADS alone
Design
Model development and validation study
Follow-up
30 days
Key result
A multimodal deep learning system integrating automated CAD-RADS, anatomy, and demographics predicted 30-day MACE with AUROC 0.82, outperforming RCRI and human CAD-RADS.
Authors
Loading...
May improve MACE prediction before elective noncardiac surgery; extends RCRI and CAD-RADS but should not yet change practice.
Does a fully automated multimodal deep learning system integrating CCTA and clinical data improve 30-day MACE prediction compared to RCRI and CAD-RADS alone in patients undergoing elective non-cardiac surgery?
A fully automated deep learning system combining CCTA imaging features with clinical data significantly improves 30-day perioperative MACE prediction compared to standard clinical risk scores.
Lu et al. (2026) studied this question. A multimodal deep learning system integrating automated CAD-RADS, anatomy, and demographics predicted 30-day MACE with AUROC 0.82, outperforming RCRI and human CAD-RADS.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: