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.

Epilepsy is a major complication of stroke. We aimed to establish whether there is an association between intravenous thrombolysis, intra-arterial thrombolysis and post stroke seizure (PSS) development. Improved understanding of the relationship between reperfusion therapies and seizure development may improve post-stroke monitoring and follow-up. This was a retrospective, multicentre cohort study conducted at the Royal Melbourne Hospital and Jingling Hospital Nanjing. We included patients with anterior circulation ischemic stroke admitted 2008–2015. Patients were divided into four treatment groups 1. IV-tPA only, 2. Intra-arterial therapies (IAT) only, 3. IAT + IV-tPA and 4. stroke unit care only (i.e. no IV-tPA or IAT). To assess the association between type of reperfusion treatment and seizure incidence we used multivariable logistic regression models adjusted for age, stroke severity, 3-month functional outcome and prognostic factors. There were 1375 stroke unit care-only patients, of whom 28 (2%) developed PSS. There were 363 patients who received only IV-tPA, of whom 21 (5.8%) developed PSS. There were 93 patients who received IAT only, of whom 12 (12.9%) developed PSS and 112 that received both IV-tPA + IAT, of which 5 (4.5%) developed PSS. All reperfusion treatments were associated with seizure development compared to stroke unit care-only patients: IV-tPA only adjusted odds ratio (aOR) 3.7, 95%CI 1.8–7.4, p < 0.0001; IAT aOR 5.5, 95%CI 2.1–14.3, p < 0.0001, IAT + IV-tPA aOR 3.4, 95% CI 0.98–11.8, p = 0.05. These aORs did not differ significantly between treatment groups (IV-tPA + IAT versus IV-tPA p = 0.89, IV-tPA + IAT versus IAT, p = 0.44). Patients receiving thrombolytic or intra-arterial reperfusion therapies for acute ischemic stroke are at higher risk of epilepsy and may benefit from longer follow-up. No evidence for an additive or synergistic effect of treatment modality on seizure development was found.

Background: Ischemic stroke is a leading cause of new-onset seizures. Cortical ischemia and large ischemic lesion size are among the most consistently reported risk factors for post-stroke seizures. Alberta Stroke Program Early CT Score (ASPECTS) is a simple and reliable tool for quantifying the extent of cerebral ischemia and may function as a screening tool for patients with high risk of seizure development. We investigated the association of post-stroke seizures with the extent of ischemia assessed by ASPECTS and with cortical involvement identified on non-contrast CT (NCCT). Methods: This cohort study was based on a prospectively maintained clinical database of acute ischemic stroke patients who were given intravenous tissue plasminogen activator treatment. We included patients with anterior circulation stroke admitted between January 2008 and October 2014. Patients with pre-stroke seizures were excluded. Clinical data and seizure follow-up data were collected. NCCT scans acquired both on stroke admission and at 24 h were analyzed. Logistic regression and cox regression were performed in statistical analysis. Results: A total of 348 patients (median age 73 years, interquartile range [IQR] 63-80, 55% male) were included. During follow-up (median duration 559 days, IQR 107.5-1188.5 days), 22 (6.3%) patients developed post-stroke seizures. Median time from stroke to seizure onset was 138 days (IQR 10-342 days). In univariate logistic regression, both ASPECTS on admission (OR 0.69 per 1-point increase; 95% CI 0.55-0.86; p = 0.001) and at 24 h (OR 0.80 per 1-point increase; 95% CI 0.70-0.92; p = 0.002) were significantly associated with post-stroke seizures. Cortical involvement at 24 h also correlated with seizure occurrence (OR 3.01; 95% CI 1.08-8.34; p = 0.03). Cox regression confirmed the higher risk of developing seizures at any time point in patients with lower ASPECTS value and cortical ischemia. Of note, ASPECTS was the only independent predictor for post-stroke seizures in multivariate logistic regression. Conclusion: The extent of ischemia assessed by ASPECTS and cortical involvement identified on NCCT were associated with the development of post-stroke seizures.