Epilepsy is a complex and widespread neurological disorder that has become a global public health issue. In recent years, significant progress has been made in the use of wearable devices for seizure monitoring, prediction, and treatment. This paper reviewed the applications of invasive and non-invasive wearable devices in seizure monitoring, such as subcutaneous EEG, ear-EEG, and multimodal sensors, highlighting their advantages in improving the accuracy of seizure recording. It also discussed the latest advances in the prediction and treatment of seizure using wearable devices.
Conducting research on patient-specific electroencephalography-based epilepsy seizure prediction methods enables early identification of seizure risk, providing a basis for timely intervention and treatment. However, existing methods fail to simultaneously account for the dynamic temporal feature differences of electroencephalography signals and the spatial correlations between leads when representing spatio-temporal features, limiting the representation of preictal electroencephalography features and consequently affects prediction performance. To address this issue, this paper proposes a patient-specific electroencephalography seizure prediction method based on global dynamic multi-scale spatio-temporal features. By designing a dynamic temporal attention (DTA) branch, it captures instantaneous dynamic features through convolutional extraction of feature differences between adjacent sampling points, and by designing a multi-scale spatial attention (MSSA) branch, it represents multi-scale spatial features among channels using receptive fields of convolution kernels of different sizes. Furthermore, considering the limited local receptive field of convolution operations, attention modules are introduced into the aforementioned branches to represent global information. Finally, a feature fusion (FF) branch is used to represent global dynamic multi-scale spatio-temporal features, aiming to achieve high-precision epilepsy seizure prediction. The accuracy on two public epilepsy electroencephalography datasets reached 95.36% and 72.98%, with sensitivities of 94.08% and 66.40%, and specificities of 96.91% and 79.55%, respectively. Experimental results indicate that the proposed global dynamic multi-scale spatio-temporal features can effectively characterize the dynamic temporal variations and inter-channel spatial correlations of electroencephalography signals, providing strong support for early warning of epileptic seizures.