Haijiao Wang

Background: We aimed to evaluate the association between epilepsy and suicidality, including suicidal ideation, attempts and completed suicide.Methods: We systematically searched PubMed, Embase, Cochrane Online Library, and Clinicaltrials.gov from 1946 to June 21, 2021 and assessed the quality of the studies using the Newcastle–Ottawa Scale. We calculated the pooled OR and the crude rate for suicidal ideation, suicide attempts and completed suicide in patients with epilepsy (PWE).Results: We screened 2,786 studies and included 88 articles with 1,178,401 PWE and 6,900,657 participants as controls. Search terms included epilepsy and suicide. The pooled rates of suicidal ideation, suicide attempts and completed suicide in PWE were 19.73% (95% CI: 17.00–22.62%), 5.96% (95% CI: 4.82–7.20%), and 0.24% (95% CI: 0.11–0.42%), respectively. Compared to the control group, PWE were at a significantly higher risk of total suicidality (pooled OR, 2.60; 95%: 2.13–3.18), including suicidal ideation (pooled OR, 2.70; 95% CI, 2.21–3.30), suicide attempts (pooled OR, 2.74; 95% CI, 2.08–3.61) and completed suicide (pooled OR, 2.36; 95% CI, 1.45–3.83). Subgroup analyses showed significant differences in the subgroups of the measurement of suicidality.Conclusion: The rate of suicidal ideation, suicide attempts and completed suicide in PWE were about 19.73, 5.96, and 0.24%. And there was an increased risk of suicidality in PWE especially temporal lobe epilepsy and drug-resistant epilepsy. Clinicians need to be aware of this risk in PWE with early identification and prevention at the time of diagnosis.Protocol Registration: PROSPERO CRD42021278220.

Background and purpose: Post‐stroke seizures (PSSs) are some of the most common complications of stroke and are associated with poor outcomes in patients. Endovascular treatment (EVT) is the standard of care for patients with acute ischaemic stroke related large‐vessel occlusion. However, whether EVT increases the risk of PSSs remains controversial; the association between PSSs and EVT is poorly understood.Methods: PubMed, Embase and the Cochrane Library were searched for relevant studies published from 1995 to 6 December 2021. The overall incidence of PSSs in patients treated with EVT and the separate incidence for all included studies in each subgroup, stratified by the type of treatment or time of onset, were calculated. The pooled odds ratio and confidence interval were calculated to quantify the effects of EVT on PSS occurrence.Results: In all, 946 studies were screened and 16 articles were included, with a total sample size of 12,664 patients; 7836 patients received EVT, of whom 460 had PSS. The pooled incidence of PSS after EVT was 5.8%, which was similar to patients treated with mechanical thrombectomy (5.3%), intra‐arterial thrombolysis (6.8%) or bridging therapy (5.4%). The cumulative incidence of post‐stroke epilepsy (6.0%) was almost twice that of acute symptomatic seizures (3.6%). The pooled odds ratio for the relationship between EVT and PSS was 1.91 (95% confidence interval 0.98–3.73).Conclusions: The cumulative incidence of stroke patients treated with EVT who developed seizures was 5.8%, and EVT was non‐significantly associated with the occurrence of seizures after stroke.

Objectives: To explore the trend of sudden unexpected death in epilepsy (SUDEP) incidence rate over time in rural west China.Methods: We scanned probable SUDEP patients from the epilepsy program between 2010 and 2019 in rural West China and performed a verbal autopsy for each eligible patient. We calculated the crude and sex-adjusted incidence rate of SUDEP per person-year over a calendar year and the year of follow-up. We calculated the incidence rate ratio with the Poisson model in STATA 12.0 and calculated the annual percentage change (APC) and average annual percentage change in Joinpoint Trend Analysis Software 4.8.0.1 to analyze the trend of SUDEP incidence rate.Results: In 2010–2019, 44 probable SUDEPs were identified from 10,128 patients with a total person-year of 31,347. The crude and sex-adjusted incidence rates of SUDEP were 1.40 and 1.45%0. Twenty-five (56.8%) of the 44 probable SUDEPs had no generalized tonic-clonic seizure 3 months before their death. The incidence of probable SUDEP decreased significantly in the calendar year [APC = −11.7, 95% confidence interval (CI): −21.7 to −0.3] and in time of follow-up (average annual percentage change = −21.2, 95% CI: −34.3 to −5.4). Comparing the first 5 years in follow-up with the subsequent 3 years, the incidence rate of SUDEP decreased significantly (estimated incidence rate ratio = 0.4, 95% CI: 0.2 to 0.8).Significance: SUDEP happened to 1.4 cases per thousand patient-years in convulsive epilepsy in rural west China between 2010 and 2019. The incidence rate of SUDEP presented a downward trend over the time of follow-up.

Background: Posttraumatic epilepsy (PTE) is a common sequela after traumatic brain injury (TBI), and identifying high-risk patients with PTE is necessary for their better treatment. Although artificial neural network (ANN) prediction models have been reported and are superior to traditional models, the ANN prediction model for PTE is lacking. Objective: We aim to train and validate an ANN model to anticipate the risks of PTE. Methods: The training cohort was TBI patients registered at West China Hospital. We used a 5-fold cross-validation approach to train and test the ANN model to avoid overfitting; 21 independent variables were used as input neurons in the ANN models, using a back-propagation algorithm to minimize the loss function. Finally, we obtained sensitivity, specificity, and accuracy of each ANN model from the 5 rounds of cross-validation and compared the accuracy with a nomogram prediction model built in our previous work based on the same population. In addition, we evaluated the performance of the model using patients registered at Chengdu Shang Jin Nan Fu Hospital (testing cohort 1) and Sichuan Provincial People’s Hospital (testing cohort 2) between January 1, 2013, and March 1, 2015. Results: For the training cohort, we enrolled 1301 TBI patients from January 1, 2011, to December 31, 2017. The prevalence of PTE was 12.8% (166/1301, 95% CI 10.9%-14.6%). Of the TBI patients registered in testing cohort 1, PTE prevalence was 10.5% (44/421, 95% CI 7.5%-13.4%). Of the TBI patients registered in testing cohort 2, PTE prevalence was 6.1% (25/413, 95% CI 3.7%-8.4%). The results of the ANN model show that, the area under the receiver operating characteristic curve in the training cohort was 0.907 (95% CI 0.889-0.924), testing cohort 1 was 0.867 (95% CI 0.842-0.893), and testing cohort 2 was 0.859 (95% CI 0.826-0.890). Second, the average accuracy of the training cohort was 0.557 (95% CI 0.510-0.620), with 0.470 (95% CI 0.414-0.526) in testing cohort 1 and 0.344 (95% CI 0.287-0.401) in testing cohort 2. In addition, sensitivity, specificity, positive predictive values and negative predictors in the training cohort (testing cohort 1 and testing cohort 2) were 0.80 (0.83 and 0.80), 0.86 (0.80 and 0.84), 91% (85% and 78%), and 86% (80% and 83%), respectively. When calibrating this ANN model, Brier scored 0.121 in testing cohort 1 and 0.127 in testing cohort 2. Compared with the nomogram model, the ANN prediction model had a higher accuracy (P=.01). Conclusions: This study shows that the ANN model can predict the risk of PTE and is superior to the risk estimated based on traditional statistical methods. However, the calibration of the model is a bit poor, and we need to calibrate it on a large sample size set and further improve the model.

Objective To characterize the features of thalamocortical functional connectivity during seizure recurrence at the time of antiseizure medication (ASM) withdrawal. Methods Patients with chronic epilepsy who attempted to discontinue medications were prospectively registered and followed up; 19 patients remained seizure-free (SF-group), 18 patients had seizure relapses (SR-group) after ASM withdrawal, and 28 healthy controls were recruited. Resting-state functional magnetic resonance imaging was performed before ASM withdrawal. Thalamus subdivisions were set as seeds to calculate voxelwise functional connectivity. Partial correlation analysis between functional connectivity and clinical variables was performed. A support vector machine was used to assess the predictive ability of the specific functional connectivity for seizure relapse. Results The within-group comparison indicated that the SR-group had more extensive functional connectivity than the SF-group; the left inferior pulvinar, left medial pulvinar, and right anterior pulvinar showed a significantly stronger functional connection with the precuneus in the SR-group than in the SF-group (Gaussian random field correction, voxel-level p < .001 and cluster-level p < .05). In the SR-group, a positive correlation was found between the left inferior pulvinar–precuneus connectivity and the active period (r = .46, p = .05), seizure-free period (r = .67, p = .002), and disease duration (r = .53, p = .02), and between the left medial pulvinar–precuneus connectivity and the seizure-free period (r = .58, p = .01). The combination of these thalamocortical connections showed a high predictive ability, with an area under the curve of .92 and accuracy of .90 (p = .01). Significance This study determined distinct features of thalamocortical functional connectivity at the time of ASM withdrawal in patients with and without seizure relapse, showing a potential for predicting seizure outcomes following ASM withdrawal.

BACKGROUND Posttraumatic epilepsy (PTE) is a common sequela after traumatic brain injury (TBI), and identifying high-risk patients with PTE is necessary for their better treatment. Although artificial neural network (ANN) prediction models have been reported and are superior to traditional models, the ANN prediction model for PTE is lacking. OBJECTIVE We aim to train and validate an ANN model to anticipate the risks of PTE. METHODS The training cohort was TBI patients registered at West China Hospital. We used a 5-fold cross-validation approach to train and test the ANN model to avoid overfitting; 21 independent variables were used as input neurons in the ANN models, using a back-propagation algorithm to minimize the loss function. Finally, we obtained sensitivity, specificity, and accuracy of each ANN model from the 5 rounds of cross-validation and compared the accuracy with a nomogram prediction model built in our previous work based on the same population. In addition, we evaluated the performance of the model using patients registered at Chengdu Shang Jin Nan Fu Hospital (testing cohort 1) and Sichuan Provincial People’s Hospital (testing cohort 2) between January 1, 2013, and March 1, 2015. RESULTS For the training cohort, we enrolled 1301 TBI patients from January 1, 2011, to December 31, 2017. The prevalence of PTE was 12.8% (166/1301, 95% CI 10.9%-14.6%). Of the TBI patients registered in testing cohort 1, PTE prevalence was 10.5% (44/421, 95% CI 7.5%-13.4%). Of the TBI patients registered in testing cohort 2, PTE prevalence was 6.1% (25/413, 95% CI 3.7%-8.4%). The results of the ANN model show that, the area under the receiver operating characteristic curve in the training cohort was 0.907 (95% CI 0.889-0.924), testing cohort 1 was 0.867 (95% CI 0.842-0.893), and testing cohort 2 was 0.859 (95% CI 0.826-0.890). Second, the average accuracy of the training cohort was 0.557 (95% CI 0.510-0.620), with 0.470 (95% CI 0.414-0.526) in testing cohort 1 and 0.344 (95% CI 0.287-0.401) in testing cohort 2. In addition, sensitivity, specificity, positive predictive values and negative predictors in the training cohort (testing cohort 1 and testing cohort 2) were 0.80 (0.83 and 0.80), 0.86 (0.80 and 0.84), 91% (85% and 78%), and 86% (80% and 83%), respectively. When calibrating this ANN model, Brier scored 0.121 in testing cohort 1 and 0.127 in testing cohort 2. Compared with the nomogram model, the ANN prediction model had a higher accuracy (P=.01). CONCLUSIONS This study shows that the ANN model can predict the risk of PTE and is superior to the risk estimated based on traditional statistical methods. However, the calibration of the model is a bit poor, and we need to calibrate it on a large sample size set and further improve the model.