Cross-modal autoencoder framework learns holistic representations of cardiovascular stateAdityanarayanan RadhakrishnanSam F. Friedmanet al.2023Nature Communications