Epilepsy is a disease in which patients undergo seizures caused by brain functionality disorder. Clinically, it is usually diagnosed by experienced clinicians according to continuous electroencephalography (cEEG), which is time consuming even for experienced doctors. Meanwhile, amplitude integrated electroencephalography (aEEG) has shown potential to detect epileptic seizures. Therefore, the paper proposes a hybrid seizure detection algorithm by combining cEEG-based seizure detection algorithm and aEEG-based seizure detection algorithm to detect seizures. In cEEG-based seizure detection algorithm, cEEG signals are divided into 5 s epoch with 4 s overlap and multi-domain features are extracted from each epoch. Then random forest classification is applied to do seizure detection. In aEEG-based seizure detection algorithm, morphological filter is applied to do spike detection and determine whether there are seizures after transforming the cEEG signals into aEEG signals. In order to evaluate the generality of the proposed method, experiments are performed on two independent datasets, including a publicly available EEG dataset (CHB-MIT) and an epileptic dataset collected by using the EEG device developed by the Hangzhou Neuro Science and Technology Co., Ltd. In the CHB-MIT dataset, the accuracy (AC), specificity (SP), sensitivity based on the event (SE), and false positive ratio based on the event (FPRE) obtained by the hybrid method are 99.36%, 82.98%, 99.41%, and 0.57 times/h, respectively. In the dataset we collected, the AC, SP, SE, and FPRE obtained by the hybrid method are 99.23%, 89.47%, 99.23%, and 0.71 times/h, respectively. The experimental results show that the performance of the proposed method is competitive with state-ofthe-art methods and results. Furthermore, basing on the hybrid method, this paper has developed a portable automatic seizure detection system, which can reduce the burden of clinicians in processing the large amounts of cEEG signals by detecting seizure automatically.INDEX TERMS Seizure detection, multi-domain feature, spike detection, hybrid method.

Sleep stage classification, including wakefulness (W), rapid eye movement (REM), and nonrapid eye movement (NREM) which includes three sleep stages that describe the depth of sleep, is one of the most critical steps in effective diagnosis and treatment of sleep-related disorders. Clinically, sleep staging is performed by domain experts through visual inspection of polysomnography (PSG) recordings, which is time-consuming, labor-intensive and often subjective in nature. Therefore, this study develops an automatic sleep staging system, which uses single channel electroencephalogram (EEG) signal, for convenience of wearing and less interference in the sleep, to do automatic identification of various sleep stages. To achieve the automatic sleep staging system, this study proposes a two-layer stacked ensemble model, which combines the advantages of random forest (RF) and LightGBM (LGB), where RF focuses on reducing the variance of the proposed model while LGB focuses on reducing the bias of the proposed model. Particularly, the proposed model introduces a class balance strategy to improve the N1 stage recognition rate. In order to evaluate the performance of the proposed model, experiments are performed on two datasets, including Sleep-EDF database (SEDFDB) and Sleep-EDF Expanded database (SEDFEDB). In the SEDFDB, the overall accuracy (ACC), weight F1-score (WF1), Cohen's Kappa coefficient (Kappa), sensitivity of N1 (SEN-N1) obtained by proposed model are 91.2%, 0.916, 0.864 and 72.52% respectively using subject-non-independent test (SNT). In parallel, the ACC, WF1, Kappa, SEN-N1 obtained by proposed model are 82.4%, 0.751, 0.719 and 27.15% respectively using subject-independent test (SIT). Experimental results show that the performance of the proposed model are competitive with the state-of-the-art methods and results, and the recognition rate of N1 stage is significantly improved. Moreover, in the SEDFEDB, the experimental results indicate the robustness and generality of the proposed model. INDEX TERMS Sleep stage classification, single channel EEG signal, two-layer stacked ensemble model, random forest, LightGBM.

Length of stay (LoS) in the intensive care unit (ICU) is a common outcome measure used as an indicator of both quality of care and resource use. However, the existing analysis methods of LoS are poorly interpretable and extensible, and there is controversial for the predictive performance of LoS. In this paper, the study includes data from 1,214 unplanned ICU admissions to participate in the ICU of Sichuan Provincial People's Hospital between Dec. 11, 2015 and Dec. 6, 2018. On the basis of these data, this study creates a highly accurate and predictive model using advanced preprocessing techniques, exploratory data analysis (EDA) and least absolute shrinkage and selection operator (LASSO) algorithm. Next, this study evaluates the predictive performance of the proposed model by 10-fold cross validation and external validation method using the root mean square prediction error (RMSPE), mean absolute error (MAE), and coefficient of determination (R 2). The predictive performance of the proposed model is 0.88±0.13 day for RMSPE, 0.87±0.07 day for MAE and 0.35±0.09 for R 2. Experimental results show that the performance of the proposed method are competitive with the state-of-the-art methods and results. Furthermore, this study explores the risk factors for ICU LoS in survivors and non-survivors and compare their predictive performance. INDEX TERMS Length of stay, intensive care unit, exploratory data analysis, least absolute shrinkage and selection operator.

The grading of hypoxic-ischemic encephalopathy (HIE) contributes to the clinical decision making for neonates with HIE. In this paper, an automated grading method based on electroencephalogram (EEG) data is proposed to describe the severity of HIE infants, namely mild asphyxia, moderate asphyxia and severe asphyxia. The automated grading method is based on a multi-class support vector machine (SVM) classifier, and the input features of SVM classifier include long-term features which are extracted by decomposing the EEG data into different 64 s epoch data and short-term features which are extracted by segmenting the 64 s epoch data into 8 s epoch data with 4 s overlap. Of note, the correlation coefficient and asymmetry extracted in this paper have obvious discriminating capability in HIE infants classification. The experimental results show that the proposed method can achieve the classification accuracy of 78.3%, 75.8% and 87.0% of the mild asphyxia group, moderate asphyxia group and severe asphyxia group, respectively. Moreover, the overall accuracy and kappa used to evaluate the performance of the proposed method can reach 79.5% and 0.69, respectively.