The company says its machine learning approach could help flag cardiac amyloidosis from standard 12-lead ECGs, though experts say real-world validation and accuracy will be key.