ABSTRACT
In recent years, epileptic seizure detection based on electroencephalogram (EEG) has attracted the widespread attention of the academic. However, it is difficult to collect data from epileptic seizure, and it is easy to cause over fitting phenomenon under the condition of few training data. In order to solve this problem, this paper took the CHB-MIT epilepsy EEG dataset from Boston Children's Hospital as the research object, and applied wavelet transform for data augmentation by setting different wavelet transform scale factors. In addition, by combining deep learning, ensemble learning, transfer learning and other methods, an epilepsy detection method with high accuracy for specific epilepsy patients was proposed under the condition of insufficient learning samples. In test, the wavelet transform scale factors 2, 4 and 8 were set for experimental comparison and verification. When the wavelet scale factor was 8, the average accuracy, average sensitivity and average specificity was 95.47%, 93.89% and 96.48%, respectively. Through comparative experiments with recent relevant literatures, the advantages of the proposed method were verified. Our results might provide reference for the clinical application of epilepsy detection.
Subject(s)
Child , Humans , Algorithms , Deep Learning , Electroencephalography , Epilepsy/diagnosis , Seizures/diagnosis , Signal Processing, Computer-Assisted , Wavelet AnalysisABSTRACT
Objective To investigate the predictive value of serum glycated hemoglobin A1c (HbA1c)level and early neurological deterioration (END)in patients with acute cerebral infarction. Methods From June 2014 to January 2016,the consecutive patients with acute ischemic stroke (interval time from onset to admission < 3 days)admitted to the Department of Neurology,Suqian People′s Hospital were enrolled retrospectively. The neurological deficits of all patients were evaluated on the day of admission with the National Institutes of Health Stroke Scale (NIHSS). The symptoms of neurological deficits in patients were evaluated repeatedly for all patients within 7 days after admission. Any score increased ≥2 compared with before admission was defined as END. The 226 patients met the inclusion and exclusion criteria were enrolled,and they were divided into either an END group (n = 50,22. 1%)or a non-END group (n = 176,77. 9%)according to whether END occurred after acute cerebral infarction. Univariate analysis was used to analyze the differences between the 2 groups. Multivariable Logistic regression analysis was used to analyze the correlation between the HbA1c level and END. Results Compared with the non-END group,there were significant differences in the age,prevalence of diabetes,NIHSS score,hypersensitive C-reactive protein level,rate of ASPECT score 0-7,and serum HbAlc level in the patients of the END group (all P < 0. 05). The results of logistic regression analysis showed that the elevated hypersensitive C-reaction protein and serum HbA1c levels were the independent risk factors for END of acute cerebral infarction (OR,1. 048 and 1. 809 respectively,95% CI 1. 008 -1. 089 and 1. 429 -2. 292 respectively;P = 0. 018 and 0. 002 respectively). Conclusion The increased serum HbA1c level is an independent risk factor for END of acute cerebral infarction. It has certain predictive value for END.
ABSTRACT
Objective To investigate the correlation between ankle-brachial index (ABI) and early neurological deterioration (END) in patients with acute ischemic stroke. Methods Consecutive patients with acute ischemic stroke admitted to hospital within 7 d after onset from January 2014 to December 2014 were enroled. Bidirectional Doppler flow detector was used to detect ABI. END was defined as the increased National Institutes of Health Stroke Scale score ≥2 or the increased motor score ≥1. The demographic characteristics, vascular risk factors, laboratory parameters, and the incidence of END were identified and analyzed. Results A total of 210 patients with acute ischemic stroke were enroled, including 51 had END and 159 did not have END. Univariate analysis showed that the proportion of patient with ABI ≤0. 9 of the END group was significantly higher than that of the non-END group (43. 1% vs. 22. 0% ; χ2 = 8. 714, P =0. 003). Multivariable logistic regression analysis showed that ABI ≤0. 9 (odds ratio 2. 688, 95% confidence interval 1. 265 - 5. 052; P = 0. 009) was independently associated with END in patients with ischemic stroke after adjusting for the confounding factors, such as age, sex, baseline systolic blood pressure, and ischemic heart disease. Conclusion The low ABI was associated with the occurrence of END in patients with acute ischemic stroke.