Search

• We provide two models for predicting drug-resistant epilepsy in patients with newly-diagnosed epilepsy. • EEG features extracting from deep learning and clinical features are used to predict the response to anti-seizure medications. • Deep-learning approaches will help to develop personalized models for predicting drug-resistant epilepsy. 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. 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. 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. 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. [ABSTRACT FROM AUTHOR]

Objective: To establish and validate a predictive formula for calculating the possibility of developing delayed neurological sequelae (DNS) after acute carbon monoxide (CO) poisoning to facilitate better decision-making about treatment strategies.Methods: This study retrospectively enrolled 605 consecutive patients who had been newly diagnosed with CO poisoning from the Central Hospital of Enshi Prefecture between January 1, 2015 and December 31, 2020. The cohort was randomly divided into two subgroups: the development cohort (n = 104) and validation cohort (n = 44). Univariate analysis and backward elimination of multivariate logistic regression were used to identify predictive factors, and a predictive formula was established. The performance was assessed using the area under the curve (AUC), the mean AUC of five-fold cross-validation, and calibration plots.Results: The formula included four commonly available predictors: initial GCS score, duration of exposure, CK, and abnormal findings on MRI. We next created a formula to calculate the risk score for developing DNS: Risk score = -4.54 + 3.35 * (Abnormal findings on MRI = yes) - 0.51 * (Initial GCS score) + 0.65 * (Duration of exposure) + 0.01 * (CK). Then, the probability of developing DNS could be calculated: Probability of DNS = 1/(1 + e Risk score). The model revealed good discrimination with AUC, and mean AUC of fivefold cross-validation in two cohort, and the calibration plots showed good calibration.Conclusions: This study established a prediction predictive formula for predicting developing of DNS, which could facilitate better decision-making about treatment strategies. [ABSTRACT FROM AUTHOR]