
The study evaluated a deep learning model's sleep scoring on EEG and wearable data from 50 epilepsy patients across 223 nights. Accuracy and Cohen’s kappa were moderate (EEG: 0.73/0.59; wearable: 0.61/0.43), with low sensitivity in stage N1. Patients with seizures slept longer and had more N2 sleep. Wearable-based sleep monitoring is feasible, but model accuracy must improve for clinical use.
Like
Save
Share