Transferring a deep learning model from healthy subjects to stroke patients in a motor imagery brain-computer interface.

Transfer learning from healthy to stroke can enhance the clinical use of BCI algorithms by overcoming the challenge of insufficient clinical data for optimal training.