Accurate early prediction of epileptic seizures can provide timely treatment for patients. Previous studies have mainly focused on a single temporal or spatial dimension, making it difficult to take both relationships into account. Therefore, the effective properties of electroencephalograms (EEGs) may not be fully evaluated. To solve this problem, we propose a spatiotemporal graph attention network (STGAT) based on synchronization. The spatial and functional connectivity information between EEG channels was extracted by using the phase locking values (PLVs) first, which allowed multichannel EEG signals to be modeled as graph signals. Afterward, the STGAT model was used to dynamically learn the temporal correlation properties of EEG sequences and explore the spatial topological structure information of multiple channels. Experimental results demonstrated that the STGAT model was able to obtain spatiotemporal correlations and achieve good results on two benchmark datasets. The accuracy, specificity and sensitivity were 98.74%, 99.21% and 98.87%, respectively, on the CHB-MIT dataset. Moreover, all evaluation indices of the private dataset had reached more than 98.8%, with the area under the curve (AUC) reaching 99.96% . The proposed method is superior or comparable to the state-of-the-art models. Extensive experiments demonstrate that our end-to-end automatic seizure prediction model can be extended to design clinical assistant decision systems.

Background: Epilepsy is one of the diseases of the nervous system, which has a large population in the world. Traditional diagnosis methods mostly depended on the professional neurologists’ reading of the electroencephalogram (EEG), which was time-consuming, inefficient, and subjective. In recent years, automatic epilepsy diagnosis of EEG by deep learning had attracted more and more attention. But the potential of deep neural networks in seizure detection had not been fully developed. Methods: In this article, we used a one-dimensional convolutional neural network (1-D CNN) to replace the residual network architecture’s traditional convolutional neural network (CNN). Moreover, we combined the Independent recurrent neural network (indRNN) and CNN to form a new residual network architecture-independent convolutional recurrent neural network (RCNN). Our model can achieve an automatic diagnosis of epilepsy EEG. Firstly, the important features of EEG were learned by using the residual network architecture of 1-D CNN. Then the relationship between the sequences were learned by using the recurrent neural network. Finally, the model outputted the classification results. Results: On the small sample data sets of Bonn University, our method was superior to the baseline methods and achieved 100% classification accuracy, 100% classification specificity. For the noisy real-world data, our method also exhibited powerful performance. Conclusion: The model we proposed can quickly and accurately identify the different periods of EEG in an ideal condition and the real-world condition. The model can provide automatic detection capabilities for clinical epilepsy EEG detection. We hoped to provide a positive significance for the prediction of epileptic seizures EEG.

Background: Epilepsy was defined as an abnormal brain network model disease in the latest definition. From a microscopic perspective, it is also particularly important to observe the Mutual Information (MI) of the whole brain network based on different lead positions. Methods: In this study, we selected EEG data from representative temporal lobe and frontal lobe epilepsy patients. Based on Phase Space Reconstruction and the calculation of MI indicator, we used Complex Network technology to construct a dynamic brain network function model of epilepsy seizure. At the same time, about the analysis of our network, we described the index changes and propagation paths of epilepsy discharge in different periods, and spatially monitors the seizure change process based on the analysis of the parameter characteristics of the complex network. Results: Our model portrayed the functional synergy between the various regions of the brain and the state transition during the seizure process. We also characterized the EEG synchronous propagation path and core nodes during seizures. The results shown the full node change path and the distribution of important indicators during the seizure process, which makes the state change of the seizure process more clearly. Conclusion: In this study, we have demonstrated that synchronization-based brain networks change with time and space. The EEG synchronous propagation path and core nodes during epileptic seizures can provide a reference for finding the focus area.

Characterizing the synchronous changes of epileptic seizures in different stages between different regions is profound to understand the transmission pathways of epileptic brain network and epileptogenic foci. There is currently no adequate quantitative calculation method for describing the propagation pathways of electroencephalogram (EEG) signals in the brain network from the short and long term. The goal of this study is to explore the innovative method to locate epileptic foci, mapping synchronization in the brain networks based on EEG. Mutual information was used to analyze the short-term synchronization in the full electrodes; while nonlinear dynamics quantifies the statistical independencies in the long –term among all electrodes. Then graph theory based on the complex network was employed to construct a dynamic brain network for epilepsy patients when they were awake, asleep and in seizure, analyzing the changing topology indexes. Epileptic network achieved a high degree of nonlinear synchronization compared to awake time. and the main path of epileptiform activity was revealed by searching core nodes. The core nodes of the brain network were in connection with the onset zone. Seizures always happened with a high degree of distribution. This study indicated the path of EEG synchronous propagation in seizures, and core nodes could locate the epileptic foci accurately in some epileptic patients.

Blood pressure is an important reflection of human health. With the introduction of precision medicine in recent years, due to the individual differences, the range of hypertension should not be used as a test standard for a set of data. But it should be determined by the individual's daily blood pressure values. This paper chooses the gray model, which is suitable for small-sample data prediction, complete a short-term prediction of blood pressure to determine the individual's fluctuations in blood pressure data. However, the standard gray model GM (1,1) doesn't satisfy. Firstly, coordination conditions, the background and gray values don't have the consistency of the transformation. Secondly, when the cumulative model is generated, the first point of the original sequence doesn't play a role in enhancing the accuracy of predictions; resulting in inaccurate prediction results. Therefore, we combine the improved gray model GM(1,1) with an equal-dimension information substitution model to predict the individuals blood pressure. The experimental results indicate that this model improves the accuracy of blood pressure prediction and it could prevent while treating early hypertension or other diseases, meeting the actual needs of doctors and patients.