All Competitors

Genomic foundation model training framework whose joint-embedding predictive objective learns functional representations of masked DNA, not tokens.

Joint embedding predictive architecture for ICU bedside waveforms, learning frozen ABP, ECG and PPG representations for five-minute risk estimation.

Generalist neuroimaging vision foundation model pretrained on 5.24M clinical MRI and CT volumes for radiologic diagnosis and report generation.

Self-supervised 12-lead ECG encoder that predicts masked patches in latent space, using a cross-lead attention mask shaped by clinical reading.

Joint-embedding predictive foundation model pretrained on over a million unlabeled ECGs, learning transferable 12-lead representations for diagnosis.