Pretrained Sleep EEG Brain-Health Model for Transfer Learning

45 views

Skip to first unread message

Wolfgang Ganglberger

unread,

Jul 7, 2026, 1:35:06 PM (5 days ago) Jul 7

to physionet-challenges

For teams exploring transfer learning in the 2026 PhysioNet Challenge: Philosopher’s Stone is a public pretrained sleep EEG model with a 1024-D brain-health latent space and four high-level outputs: brain health, total cognition, fluid cognition, and crystallized cognition.

It can be used as a feature extractor, transfer-learning backbone, or auxiliary representation for a Challenge-specific classifier: https://github.com/bdsp-core/philosophers-stone