The HeartTTable multimodal AI model integrating 3D CINE reconstruction and clinical/CMR data predicted 5-year MACE with a time-dependent AUC of 0.934 and Harrell's C-index of 0.897, outperforming models using only tabular data (AUC 0.772) and established risk scores.
Observational (n=4,549)
Yes
Does the HeartTTable multimodal model improve 5-year MACE risk prediction in post-PCI AMI patients compared to tabular-only models and traditional risk scores?
A multimodal AI model integrating 3D spatiotemporal CMR reconstructions with clinical data significantly improves long-term (5-year) MACE prediction in post-PCI AMI patients compared to traditional risk scores and tabular-only models.
Effect estimate: time-dependent AUC 0.934 (HeartTTable 3D + Table model) vs 0.772 (Table only model) (95% CI 95% CI 0.907–0.959)
Absolute Event Rate: 0.934% vs 0.772%
p-value: p=<0.0001
Artificial intelligence has made significant strides in predicting major adverse cardiovascular events (MACE) in patients with acute myocardial infarction (AMI) following percutaneous coronary intervention. However, most existing methods rely solely on tabular variables derived from clinical data and cardiac magnetic resonance (CMR), without fully leveraging the predictive potential of the CMR imaging modality itself. Moreover, these approaches often overlook the synergistic benefits of multimodal integration between imaging and tabular data. In addition, current models primarily focus on short-term MACE risk assessment (e.g., within 6 months or 1 year), limiting their applicability for long-term prognostication. To address these limitations, we first developed ReconSeg3D, a model that reconstructs short-axis cine CMR stacks into temporally-resolved 3D bi-ventricular volumes, capturing fine-grained cardiac anatomy and dynamic motion. These bi-ventricular sequences were then integrated with 45 clinical and CMR-derived variables using spatiotemporal decomposition and cross-attention mechanisms to construct a multimodal MACE prediction model-HeartTTable. HeartTTable achieved a 5-year time-dependent AUC of 0.934 (95% CI 0.907-0.959) and a Harrell's C-index of 0.897 for predicting MACE risk, significantly outperforming models based solely on clinical and CMR-derived tabular features, and demonstrated strong capabilities in postoperative risk stratification. Our study contributes to improved long-term postoperative management for AMI patients by offering clinicians an objective, data-driven decision-support tool.
Gao et al. (Fri,) conducted a observational in Patients with acute myocardial infarction undergoing percutaneous coronary intervention within 12 hours of symptom onset with CMR imaging done within 7 days post-procedure (n=4,549). HeartTTable AI multimodal model integrating 3D CINE cardiac reconstruction and clinical plus CMR-derived tabular data vs. Models using only clinical and CMR tabular features or only 3D CINE reconstruction-derived data; established risk scores (GRACE, Eitel, Glasgow) was evaluated on Major adverse cardiovascular events (MACE) within 5 years post-PCI, defined as composite of all-cause death, unplanned revascularization, reinfarction, hospitalization for heart failure, and malignant arrhythmia (time-dependent AUC 0.934 (HeartTTable 3D + Table model) vs 0.772 (Table only model), 95% CI 95% CI 0.907–0.959, p=<0.0001). The HeartTTable multimodal AI model integrating 3D CINE reconstruction and clinical/CMR data predicted 5-year MACE with a time-dependent AUC of 0.934 and Harrell's C-index of 0.897, outperforming models using only tabular data (AUC 0.772) and established risk scores.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: