KiT: A Deep Learning Diagnostic Model of EEG Signals from Disorders of Consciousness Patients Based on Kolmogorov Arnold Network - Vision Transformer

In recent years, deep learning models for EEG signal analysis algorithms have become popular within the field of brain-computer interfaces (BCI) due to their high accuracy, while the accurate classification of patients with chronic disorders of consciousness (DoC) is crucial in assessment and prognosis, traditional behavioral scales are widely used in clinical practice, but they are highly susceptible to subjective constraints, such as professionalism, so researchers have conducted numerous studies on the performance of determining the minimally conscious state (MCS) and the Therefore, researchers have done a lot of research on deep learning algorithms for determining the minimally conscious state (MCS) and the unconscious state (UWS/VS). The transformer model has attracted a lot of attention in the field of EEG signal processing due to its full-director sequence learning ability, however, there is still room for improvement in the expressiveness and performance of the transformer model, and the emergence of the Kolmogorov-Arnold network offers a new way to improve the ability of the transformer model. The emergence of the Kolmogorov-Arnold network provides a new idea to improve the ability of the transformer model, and after being combined with spatial feature extraction, the KiT model performs well in DOC EEG data analysis.