All Competitors

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

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

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