CogAdapt: Transferring Clinical ECG Foundation Models to Wearable Cognitive Load Assessment via Lead Adaptation
Researchers have developed CogAdapt, a framework designed to adapt existing clinical ECG foundation models for use in wearable cognitive load assessment. This is necessary because models trained on clinical data don't directly translate to wearable sensors due to differences in signal configuration and task objectives. CogAdapt utilizes a 'LeadBridge' adapter to convert 3-lead wearable signals to 12-lead representations and a 'ProFine' strategy for progressive fine-tuning, achieving improved performance on public datasets. AI
IMPACT Enables more accurate and personalized cognitive load assessment from wearable devices by leveraging pre-trained foundation models.