Development and external validation of a predictive nomogram model of posttraumatic epilepsy: A retrospective analysis

Development and validation of a clinical nomogram for predicting in-hospital mortality in patients with traumatic brain injury prehospital: A retrospective study

Objective

Traumatic brain injury (TBI) is among the leading causes of death and disability globally. Identifying and assessing the risk of in-hospital mortality in traumatic brain injury patients at an early stage is challenging. This study aimed to develop a model for predicting in-hospital mortality in TBI patients using prehospital data from China.

Methods

We retrospectively included traumatic brain injury patients who sustained injuries due to external forces and were treated by pre-hospital emergency medical services (EMS) at a tertiary hospital. Data from the pre-hospital emergency database were analyzed, including demographics, trauma mechanisms, comorbidities, vital signs, clinical symptoms, and trauma scores. Eligible patients were randomly divided into a training set (241 cases) and a validation set (104 cases) at a 7:3 ratio. Least absolute shrinkage and selection operator (LASSO) and multivariate logistic regression were employed to identify independent risk factors. Analyzed the discrimination, calibration, and net benefit of the nomogram across both groups.

Results

17.40 % (42/241) of TBI patients died in the hospital in the training set, while 18.30 % (19/104) in the validation set. After analysis, chest trauma (odds ratio [OR] = 4.556, 95 % confidence interval [CI] = 1.861-11.152, P = 0.001), vomiting (OR = 2.944, 95%CI = 1.194-7.258, P = 0.019), systolic blood pressure (OR = 0.939, 95%CI = 0.913-0.966, P < 0.001), SpO2 (OR = 0.778, 95%CI = 0.688-0.881, P < 0.001), and heart rate (OR = 1.046, 95%CI = 1.015-1.078, P = 0.003) were identified as independent risk factors for in-hospital mortality in TBI patients. The nomogram based on the five factors demonstrated well-predictive power, with an area under the curve (AUC) of 0.881 in the training set and 0.866 in the validation set. The calibration curve and decision curve analysis showed that the predictive model exhibited good consistency and covered a wide range of threshold probabilities in both sets.

Conclusion

The nomogram based on prehospital data demonstrated well-predictive performance for in-hospital mortality in TBI patients, helping prehospital emergency physicians identify and assess severe TBI patients earlier, thereby improving the efficiency of prehospital emergency care.

Prognostic Implications of Early Prediction in Posttraumatic Epilepsy

Posttraumatic epilepsy (PTE) is a complication of traumatic brain injury that can increase morbidity, but predicting which patients may develop PTE remains a challenge. Much work has been done to identify a variety of risk factors and biomarkers, or a combination thereof, for patients at highest risk of PTE. However, several issues have hampered progress toward fully adapted PTE models. Such issues include the need for models that are well-validated, cost-effective, and account for competing outcomes like death. Additionally, while an accurate PTE prediction model can provide quantitative prognostic information, how such information is communicated to inform shared decision-making and treatment strategies requires consideration of an individual patient's clinical trajectory and unique values, especially given the current absence of direct anti-epileptogenic treatments. Future work exploring approaches integrating individualized communication of prediction model results are needed.

Insights into epileptogenesis from post-traumatic epilepsy

Post-traumatic epilepsy (PTE) accounts for 5% of all epilepsies. The incidence of PTE after traumatic brain injury (TBI) depends on the severity of injury, approaching one in three in groups with the most severe injuries. The repeated seizures that characterize PTE impair neurological recovery and increase the risk of poor outcomes after TBI. Given this high risk of recurrent seizures and the relatively short latency period for their development after injury, PTE serves as a model disease to understand human epileptogenesis and trial novel anti-epileptogenic therapies. Epileptogenesis is the process whereby previously normal brain tissue becomes prone to recurrent abnormal electrical activity, ultimately resulting in seizures. In this Review, we describe the clinical course of PTE and highlight promising research into epileptogenesis and treatment using animal models of PTE. Clinical, imaging, EEG and fluid biomarkers are being developed to aid the identification of patients at high risk of PTE who might benefit from anti-epileptogenic therapies. Studies in preclinical models of PTE have identified tractable pathways and novel therapeutic strategies that can potentially prevent epilepsy, which remain to be validated in humans. In addition to improving outcomes after TBI, advances in PTE research are likely to provide therapeutic insights that are relevant to all epilepsies.

Early prediction of drug-resistant epilepsy using clinical and EEG features based on convolutional neural network

OBJECTIVE

Machine learning utilization in electroencephalogram (EEG) analysis and epilepsy care is fast evolving. Thus, we aim to develop and validate two one-dimensional convolutional neural network (CNN) algorithms for predicting drug-resistant epilepsy (DRE) in patients with newly-diagnosed epilepsy based on EEG and clinical features.

METHODS

We included a total of 1010 EEG signal epochs and 15 clinical features from 101 patients with epilepsy. Each patient had 10 epochs of EEG signal data, with each signal recorded for 90 s. The ratio of development set and validation set was 80:20, and ten-fold cross validation was performed. First, a CNN algorithm was used to extract EEG features automatically. Then, Two one-dimensional CNNs were crafted.. Accuracy, specificity, precision, sensitivity, F1-score, kappa statistics, mean square error (MSE) and area under the curve (AUC) were calculated to evaluate the classifiers performance.

RESULTS

The clinical-EEG model showed good performance and clinical practical value, with the accuracy, specificity, precision, sensitivity, F1-score, kappa statistics, best MSE and AUC in test set were 0.99, 0.72, 0.82, 0.96, 0.89, 0.83, 32.00, 0.81, respectively, and the accuracy in validation set was 0.84. In the EEG model, the accuracy, specificity, precision, sensitivity, F1-score, kappa statistics, best MSE and AUC in test set were 0.99, 0.59, 0.82, 0.90, 0.86, 0.72, 181.76, 0.76, respectively, and the accuracy in validation set was 0.81.

CONCLUSION

We constructed a clinical-EEG model showed good potential for predicting DRE in patients with newly-diagnosed epilepsy, which could help identify patients at high risk of developing DRE at earlier stages.

Efficiency of surgery on posttraumatic epilepsy: a systematic review and meta-analysis

Posttraumatic epilepsy (PTE) accounts for approximately 20% of structural epilepsy, and surgical intervention may be a potential treatment option for these patients. Therefore, the purpose of this meta-analysis is to evaluate the effectiveness of surgical interventions for the management of PTE. Four electronic databases (Pubmed, Embase, Scopus and Cochrane library) were searched to identify studies on surgical management of PTE. Seizures reduction rate were analyzed quantitatively in a meta-analysis. Fourteen studies involving 430 PTE patients were selected for analysis, out of which 12 reported on resective surgery (RS), 2 on vagus nerve stimulation (VNS), and 2 of the 12 RS studies reported that 14 patients underwent VNS. The seizure reduction rate for surgical interventions (both RS and VNS) was 77.1% (95% confidence interval [CI]: 69.8%-83.7%) with moderate heterogeneity (I² = 58.59%, P_hetero = 0.003). Subgroup analysis based on different follow-up times revealed that the seizure reduction rate was 79.4% (95% CI: 69.1%-88.2%) within 5 years and 71.9% (95% CI: 64.5%-78.8%) beyond 5 years. The seizure reduction rate for RS alone was 79.9% (95% CI: 70.3%-88.2%) with high heterogeneity (I² = 69.85%, P_hetero = 0.001). Subgroup analysis showed that the seizure reduction rate was 77.9% (95% CI: 66%-88.1%) within 5 years and 85.6% (95% CI: 62.4%-99.2%) beyond 5 years, with 89.9% (95% CI: 79.2%-97.5%) for temporal lobectomy and 84% (95% CI: 68.2%-95.9%) for extratemporal lobectomy. The seizure reduction rate for VNS alone was 54.5% (95% CI: 31.6%-77.4%). Surgical interventions appeared to be effective for PTE patients without severe complications, RS seemed more beneficial than VNS, while temporal lobectomy is more favorable than extratemporal resection. However, further studies with long-term follow-up data are needed to better understand the relationship between VNS and PTE.

Establishment and validation of PTE prediction model in patients with cerebral contusion

Post-traumatic epilepsy (PTE) is an important cause of poor prognosis in patients with cerebral contusions. The primary purpose of this study is to evaluate the high-risk factors of PTE by summarizing and analyzing the baseline data, laboratory examination, and imaging features of patients with a cerebral contusion, and then developing a Nomogram prediction model and validating it. This study included 457 patients diagnosed with cerebral contusion who met the inclusion criteria from November 2016 to November 2019 at the Qinghai Provincial People's Hospital. All patients were assessed for seizure activity seven days after injury. Univariate analysis was used to determine the risk factors for PTE. Significant risk factors in univariate analysis were selected for binary logistic regression analysis. P < 0.05 was statistically significant. Based on the binary logistic regression analysis results, the prediction scoring system of PTE is established by Nomogram, and the line chart model is drawn. Finally, external validation was performed on 457 participants to assess its performance. Univariate and binary logistic regression analyses were performed using SPSS software, and the independent predictors significantly associated with PTE were screened as Contusion site, Chronic alcohol use, Contusion volume, Skull fracture, Subdural hematoma (SDH), Glasgow coma scale (GCS) score, and Non late post-traumatic seizure (Non-LPTS). Based on this, a Nomogram model was developed. The prediction accuracy of our scoring system was C-index = 98.29%. The confidence interval of the C-index was 97.28% ~ 99.30%. Internal validation showed that the calibration plot of this model was close to the ideal line. This study developed and verified a highly accurate Nomogram model, which can be used to individualize PTE prediction in patients with a cerebral contusion. It can identify individuals at high risk of PTE and help us pay attention to prevention in advance. The model has a low cost and is easy to be popularized in the clinic. This model still has some limitations and deficiencies, which need to be verified and improved by future large-sample and multicenter prospective studies.

An Artificial Neural Network Prediction Model for Posttraumatic Epilepsy: Retrospective Cohort Study

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.