AI can now screen for a deadly, underdiagnosed condition using data you already have.
ATTR-CM (transthyretin cardiac amyloidosis) is massively underdiagnosed in aortic stenosis patients — even those going to TAVR. Cardio Amyloid-AI changes that. The model fuses:
3D chest CT
Echo + ECG
Patient demographics → Into a single deep learning pipeline, validated against PYP nuclear imaging (the gold standard). Performance (AUC): • Multimodal model: ~0.83 • Echo-ECG model: ~0.72 • CT alone: ~0.68 • ECG alone: ~0.52 Multimodal wins — by a lot. The kicker: the model can flag ATTR-CM YEARS before formal diagnosis using retrospective data back to 2004. No extra tests. No extra cost. Just smarter use of what’s already there. ATTR-CM is treatable (tafamidis, acoramidis). Finding it earlier = better outcomes.