Epileptic Seizure Detection with an End-to-End Temporal Convolutional Network and Bidirectional Long Short-Term Memory Model

Automatic seizure detection plays a key role in assisting clinicians for rapid diagnosis and treatment of epilepsy. In view of the parallelism of temporal convolutional network (TCN) and the capability of bidirectional long short-term memory (BiLSTM) in mining the long-range dependency of multi-channel time-series, we propose an automatic seizure detection method with a novel end-to-end TCN-BiLSTM model in this work. First, raw EEG is filtered with a 0.5-45 Hz band-pass filter, and the filtered data are input into the proposed TCN-BiLSTM network for feature extraction and classification. Post-processing process including moving average filtering, thresholding and collar technique is then employed to further improve the detection performance. The method was evaluated on two EEG database. On the CHB-MIT scalp EEG database, our method achieved a segment-based sensitivity of 94.31%, specificity of 97.13%, and accuracy of 97.09%. Meanwhile, an event-based sensitivity of 96.48% and an average false detection rate (FDR) of 0.38/h were obtained. On the SH-SDU database we collected, the segment-based sensitivity of 94.99%, specificity of 93.25%, and accuracy of 93.27% were achieved. In addition, an event-based sensitivity of 99.35% and a false detection rate of 0.54/h were yielded. The total detection time consumed for 1[Formula: see text]h EEG data was 5.65[Formula: see text]s. These results demonstrate the superiority and promising potential of the proposed method in real-time monitoring of epileptic seizures.

ECG data enhancement method using generate adversarial networks based on Bi-LSTM and CBAM

Objective.The classification performance of electrocardiogram (ECG) classification algorithms is easily affected by data imbalance, which often leads to poor model prediction performance for a few classes and a consequent decrease in the overall performance of the model.Approach.To address this problem, this paper proposed an ECG data augmentation method based on a generative adversarial network (GAN) that combines bidirectional long short-term memory (Bi-LSTM) networks and convolutional block attention mechanism (CBAM) to improve the overall performance of ECG classification models. In this paper, we used two ECG databases, namely the MIT-BIH arrhythmia (MIT-BIH-AR) database and the Chinese cardiovascular disease database (CCDD). The quality of the ECG signals produced by the generated models was assessed using the percent relative difference, root mean square error, Frechet distance, dynamic time warping (DTW), and Pearson correlation metrics. In addition, we also validated the impact of our proposed data augmentation method on ECG classification performance on MIT-BIH-AR database and CCDD.Main results.On the MIT-BIH-AR database, the overall accuracy of the data-enhanced balanced dataset was improved to 99.46% for 15 types of heartbeat classification task. On the CCDD, which focuses on the detection of ventricular precession (PVC), the overall accuracy of PVC detection improved to 99.15% after performing data enhancement.Significance.The experimental results indicate that the data augmentation method proposed in this paper can further improve the ECG classification performance.

Bidirectional Long Short-Term Memory Model of SoH Prediction for Gelled-Electrolyte Batteries under Charging Conditions

The impact of different charging currents and surrounding temperatures has always been an important aspect of battery lifetime for various electric vehicles and energy storage equipment. This paper proposes a bidirectional long short-term memory model to quantify these impacts on the aging of gel batteries and calculate their state of health. The training data set of the bidirectional long short-term memory model is collected by charging and discharging the gel battery for 300 cycles in a temperature-controlled box and an automated charge and discharge device under different operating conditions. The testing set is generated by a small energy storage device equipped with small solar panels. Data for 220 cycles at different temperatures and charging currents were collected during the experiment. The results show that the mean absolute error (MAE) and root-mean-square error (RMSE) between the training set and testing set are 0.0133 and 0.0251, respectively. In addition to the proposed model providing high accuracy, the gel battery proved to be stable and long-lasting, which makes the gel battery an ideal energy storage solution for renewable energy.

ACP-BC: A Model for Accurate Identification of Anticancer Peptides Based on Fusion Features of Bidirectional Long Short-Term Memory and Chemically Derived Information

Anticancer peptides (ACPs) have been proven to possess potent anticancer activities. Although computational methods have emerged for rapid ACPs identification, their accuracy still needs improvement. In this study, we propose a model called ACP-BC, a three-channel end-to-end model that utilizes various combinations of data augmentation techniques. In the first channel, features are extracted from the raw sequence using a bidirectional long short-term memory network. In the second channel, the entire sequence is converted into a chemical molecular formula, which is further simplified using Simplified Molecular Input Line Entry System notation to obtain deep abstract features through a bidirectional encoder representation transformer (BERT). In the third channel, we manually selected four effective features according to dipeptide composition, binary profile feature, k-mer sparse matrix, and pseudo amino acid composition. Notably, the application of chemical BERT in predicting ACPs is novel and successfully integrated into our model. To validate the performance of our model, we selected two benchmark datasets, ACPs740 and ACPs240. ACP-BC achieved prediction accuracy with 87% and 90% on these two datasets, respectively, representing improvements of 1.3% and 7% compared to existing state-of-the-art methods on these datasets. Therefore, systematic comparative experiments have shown that the ACP-BC can effectively identify anticancer peptides.

Application of Bidirectional Long Short-Term Memory to Adaptive Streaming for Internet of Autonomous Vehicles

It is expected that interconnected networks of autonomous vehicles, especially during peak traffic, will face congestion challenges. Moreover, the existing literature lacks discussions on integrating next-generation wireless communication technologies into connected vehicular networks. Hence, this paper introduces a tailored bandwidth management algorithm for streaming applications of autonomous vehicle passengers. It leverages cutting-edge 6G wireless technology to create a network with high-speed transmission and broad coverage, ensuring smooth streaming application performance. The key features of bandwidth allocation for diverse streaming applications in this work include bandwidth relay and pre-loading of video clips assisted by vehicle-to-vehicle communication. Through simulations, this research effectively showcases the algorithm's ability to fulfill the bandwidth needs of diverse streaming applications for autonomous vehicle passengers. Specifically, during periods of peak user bandwidth demand, it notably increases the bandwidth accessible for streaming applications. On average, users experience a substantial 55% improvement in the bandwidth they can access. This validation affirms the viability and promise of the proposed approach in efficiently managing the intricate complexities of bandwidth allocation issues for streaming services within the connected autonomous vehicular networks.

DeepSTF: predicting transcription factor binding sites by interpretable deep neural networks combining sequence and shape

Precise targeting of transcription factor binding sites (TFBSs) is essential to comprehending transcriptional regulatory processes and investigating cellular function. Although several deep learning algorithms have been created to predict TFBSs, the models' intrinsic mechanisms and prediction results are difficult to explain. There is still room for improvement in prediction performance. We present DeepSTF, a unique deep-learning architecture for predicting TFBSs by integrating DNA sequence and shape profiles. We use the improved transformer encoder structure for the first time in the TFBSs prediction approach. DeepSTF extracts DNA higher-order sequence features using stacked convolutional neural networks (CNNs), whereas rich DNA shape profiles are extracted by combining improved transformer encoder structure and bidirectional long short-term memory (Bi-LSTM), and, finally, the derived higher-order sequence features and representative shape profiles are integrated into the channel dimension to achieve accurate TFBSs prediction. Experiments on 165 ENCODE chromatin immunoprecipitation sequencing (ChIP-seq) datasets show that DeepSTF considerably outperforms several state-of-the-art algorithms in predicting TFBSs, and we explain the usefulness of the transformer encoder structure and the combined strategy using sequence features and shape profiles in capturing multiple dependencies and learning essential features. In addition, this paper examines the significance of DNA shape features predicting TFBSs. The source code of DeepSTF is available at https://github.com/YuBinLab-QUST/DeepSTF/.

An optimized model for network intrusion detection systems in industry 4.0 using XAI based Bi-LSTM framework

Industry 4.0 enable novel business cases, such as client-specific production, real-time monitoring of process condition and progress, independent decision making and remote maintenance, to name a few. However, they are more susceptible to a broad range of cyber threats because of limited resources and heterogeneous nature. Such risks cause financial and reputational damages for businesses, well as the theft of sensitive information. The higher level of diversity in industrial network prevents the attackers from such attacks. Therefore, to efficiently detect the intrusions, a novel intrusion detection system known as Bidirectional Long Short-Term Memory based Explainable Artificial Intelligence framework (BiLSTM-XAI) is developed. Initially, the preprocessing task using data cleaning and normalization is performed to enhance the data quality for detecting network intrusions. Subsequently, the significant features are selected from the databases using the Krill herd optimization (KHO) algorithm. The proposed BiLSTM-XAI approach provides better security and privacy inside the industry networking system by detecting intrusions very precisely. In this, we utilized SHAP and LIME explainable AI algorithms to improve interpretation of prediction results. The experimental setup is made by MATLAB 2016 software using Honeypot and NSL-KDD datasets as input. The analysis result reveals that the proposed method achieves superior performance in detecting intrusions with a classification accuracy of 98.2%.

Ensemble deep learning models for protein secondary structure prediction using bidirectional temporal convolution and bidirectional long short-term memory

Protein secondary structure prediction (PSSP) is a challenging task in computational biology. However, existing models with deep architectures are not sufficient and comprehensive for deep long-range feature extraction of long sequences. This paper proposes a novel deep learning model to improve Protein secondary structure prediction. In the model, our proposed bidirectional temporal convolutional network (BTCN) can extract the bidirectional deep local dependencies in protein sequences segmented by the sliding window technique, the bidirectional long short-term memory (BLSTM) network can extract the global interactions between residues, and our proposed multi-scale bidirectional temporal convolutional network (MSBTCN) can further capture the bidirectional multi-scale long-range features of residues while preserving the hidden layer information more comprehensively. In particular, we also propose that fusing the features of 3-state and 8-state Protein secondary structure prediction can further improve the prediction accuracy. Moreover, we also propose and compare multiple novel deep models by combining bidirectional long short-term memory with temporal convolutional network (TCN), reverse temporal convolutional network (RTCN), multi-scale temporal convolutional network (multi-scale bidirectional temporal convolutional network), bidirectional temporal convolutional network and multi-scale bidirectional temporal convolutional network, respectively. Furthermore, we demonstrate that the reverse prediction of secondary structure outperforms the forward prediction, suggesting that amino acids at later positions have a greater impact on secondary structure recognition. Experimental results on benchmark datasets including CASP10, CASP11, CASP12, CASP13, CASP14, and CB513 show that our methods achieve better prediction performance compared to five state-of-the-art methods.

Grain protein function prediction based on self-attention mechanism and bidirectional LSTM

With the development of genome sequencing technology, using computing technology to predict grain protein function has become one of the important tasks of bioinformatics. The protein data of four grains, soybean, maize, indica and japonica are selected in this experimental dataset. In this paper, a novel neural network algorithm Chemical-SA-BiLSTM is proposed for grain protein function prediction. The Chemical-SA-BiLSTM algorithm fuses the chemical properties of proteins on the basis of amino acid sequences, and combines the self-attention mechanism with the bidirectional Long Short-Term Memory network. The experimental results show that the Chemical-SA-BiLSTM algorithm is superior to other classical neural network algorithms, and can more accurately predict the protein function, which proves the effectiveness of the Chemical-SA-BiLSTM algorithm in the prediction of grain protein function. The source code of our method is available at https://github.com/HwaTong/Chemical-SA-BiLSTM.

Human activity recognition for analyzing stress behavior based on Bi-LSTM

BACKGROUND

Stress is one of the critical health factors that could be detected by Human Activity Recognition (HAR) which consists of physical and mental health. HAR can raise awareness of self-care and prevent critical situations. Recently, HAR used non-invasive wearable physiological sensors. Moreover, deep learning techniques are becoming a significant tool for analyzing health data.

OBJECTIVE

In this paper, we propose a human lifelog monitoring model for stress behavior recognition based on deep learning, which analyses stress levels during activity. The proposed approach considers activity and physiological data for recognizing physical activity and stress levels.

METHODS

To tackle these issues, we proposed a model that utilizes hand-crafted feature generation techniques compatible with a Bidirectional Long Short-Term Memory (Bi-LSTM) based method for physical activity and stress level recognition. We have used a dataset called WESAD, collected using wearable sensors for model evaluation. This dataset presented four levels of stress emotion, including baseline, amusement, stress, and meditation.

RESULTS

The following results are from the hand-crafted feature approaches compatible with the bidirectional LSTM model. The proposed model achieves an accuracy of 95.6% and an F1-score of 96.6%.

CONCLUSION

The proposed HAR model efficiently recognizes stress levels and contributes to maintaining physical and mental well-being.

Short-Term Photovoltaic Power Forecasting Based on Historical Information and Deep Learning Methods

The accurate prediction of photovoltaic (PV) power is essential for planning power systems and constructing intelligent grids. However, this has become difficult due to the intermittency and instability of PV power data. This paper introduces a deep learning framework based on 7.5 min-ahead and 15 min-ahead approaches to predict short-term PV power. Specifically, we propose a hybrid model based on singular spectrum analysis (SSA) and bidirectional long short-term memory (BiLSTM) networks with the Bayesian optimization (BO) algorithm. To begin, the SSA decomposes the PV power series into several sub-signals. Then, the BO algorithm automatically adjusts hyperparameters for the deep neural network architecture. Following that, parallel BiLSTM networks predict the value of each component. Finally, the prediction of the sub-signals is summed to generate the final prediction results. The performance of the proposed model is investigated using two datasets collected from real-world rooftop stations in eastern China. The 7.5 min-ahead predictions generated by the proposed model can reduce up to 380.51% error, and the 15 min-ahead predictions decrease by up to 296.01% error. The experimental results demonstrate the superiority of the proposed model in comparison to other forecasting methods.

Dance emotion recognition based on linear predictive Meir frequency cepstrum coefficient and bidirectional long short-term memory from robot environment

Dance emotion recognition is an important research direction of automatic speech recognition, especially in the robot environment. It is an important research content of dance emotion recognition to extract the features that best represent speech emotion and to construct an acoustic model with strong robustness and generalization. The dance emotion data set is small in size and high in dimension. The traditional recurrent neural network (RNN) has the problem of long-range dependence disappearance, and due to the focus on local information of convolutional neural network (CNN), the mining of potential relationships between frames in the input sequence is insufficient. To solve the above problems, this paper proposes a novel linear predictive Meir frequency cepstrum coefficient and bidirectional long short-term memory (LSTM) for dance emotion recognition. In this paper, the linear prediction coefficient (LPC) and Meier frequency cepstrum coefficient (MFCC) are combined to obtain a new feature, namely the linear prediction Meier frequency cepstrum coefficient (LPMFCC). Then, the combined feature obtained by combining LPMFCC with energy feature is used as the extracted dance feature. The extracted features are input into the bidirectional LSTM network for training. Finally, support vector machine (SVM) is used to classify the obtained features through the full connection layer. Finally, we conduct experiments on public data sets and obtain the better effectiveness compared with the state-of-art dance motion recognition methods.

EEG Classification of Normal and Alcoholic by Deep Learning

Alcohol dependence is a common mental disease worldwide. Excessive alcohol consumption may lead to alcoholism and many complications. In severe cases, it will lead to inhibition and paralysis of the centers of the respiratory and circulatory systems and even death. In addition, there is a lack of effective standard test procedures to detect alcoholism. EEG signals are data obtained by measuring brain changes in the cerebral cortex and can be used for the diagnosis of alcoholism. Existing diagnostic methods mainly employ machine learning techniques, which rely on human intervention to learn. In contrast, deep learning, as an end-to-end learning method, can automatically extract EEG signal features, which is more convenient. Nonetheless, there are few studies on the classification of alcohol’s EEG signals using deep learning models. Therefore, in this paper, a new deep learning method is proposed to automatically extract and classify EEG’s features. The method first adopts a multilayer discrete wavelet transform to denoise the input data. Then, the denoised data are used as input, and a convolutional neural network and bidirectional long short-term memory network are used for feature extraction. Finally, alcohol EEG signal classification is performed. The experimental results show that the method proposed in this study can be utilized to effectively diagnose patients with alcoholism, achieving a diagnostic accuracy of 99.32%, which is better than most current algorithms.

Implementing a Hand Gesture Recognition System Based on Range-Doppler Map

There have been several studies of hand gesture recognition for human–machine interfaces. In the early work, most solutions were vision-based and usually had privacy problems that make them unusable in some scenarios. To address the privacy issues, more and more research on non-vision-based hand gesture recognition techniques has been proposed. This paper proposes a dynamic hand gesture system based on 60 GHz FMCW radar that can be used for contactless device control. In this paper, we receive the radar signals of hand gestures and transform them into human-understandable domains such as range, velocity, and angle. With these signatures, we can customize our system to different scenarios. We proposed an end-to-end training deep learning model (neural network and long short-term memory), that extracts the transformed radar signals into features and classifies the extracted features into hand gesture labels. In our training data collecting effort, a camera is used only to support labeling hand gesture data. The accuracy of our model can reach 98%.

Multi-Hierarchical Fusion to Capture the Latent Invariance for Calibration-Free Brain-Computer Interfaces

Brain-computer interfaces (BCI) based motor imagery (MI) has become a research hotspot for establishing a flexible communication channel for patients with apoplexy or degenerative pathologies. Accurate decoding of motor imagery electroencephalography (MI-EEG) signals, while essential for effective BCI systems, is still challenging due to the significant noise inherent in the EEG signals and the lack of informative correlation between the signals and brain activities. The application of deep learning for EEG feature representation has been rarely investigated, nevertheless bringing improvements to the performance of motor imagery classification. This paper proposes a deep learning decoding method based on multi-hierarchical representation fusion (MHRF) on MI-EEG. It consists of a concurrent framework constructed of bidirectional LSTM (Bi-LSTM) and convolutional neural network (CNN) to fully capture the contextual correlations of MI-EEG and the spectral feature. Also, the stacked sparse autoencoder (SSAE) is employed to concentrate these two domain features into a high-level representation for cross-session and subject training guidance. The experimental analysis demonstrated the efficacy and practicality of the proposed approach using a public dataset from BCI competition IV and a private one collected by our MI task. The proposed approach can serve as a robust and competitive method to improve inter-session and inter-subject transferability, adding anticipation and prospective thoughts to the practical implementation of a calibration-free BCI system.

CAttSleepNet: Automatic End-to-End Sleep Staging Using Attention-Based Deep Neural Networks on Single-Channel EEG

Accurate sleep staging results can be used to measure sleep quality, providing a reliable basis for the prevention and diagnosis of sleep-related diseases. The key to sleep staging is the feature representation of EEG signals. Existing approaches rarely consider local features in feature extraction, and fail to distinguish the importance of critical and non-critical local features. We propose an innovative model for automatic sleep staging with single-channel EEG, named CAttSleepNet. We add an attention module to the convolutional neural network (CNN) that can learn the weights of local sequences of EEG signals by exploiting intra-epoch contextual information. Then, a two-layer bidirectional-Long Short-Term Memory (Bi-LSTM) is used to encode the global correlations of successive epochs. Therefore, the feature representations of EEG signals are enhanced by both local and global context correlation. Experimental results achieved on two real-world sleep datasets indicate that the CAttSleepNet model outperforms existing models. Moreover, ablation experiments demonstrate the validity of our proposed attention module.

American Sign Language Words Recognition of Skeletal Videos Using Processed Video Driven Multi-Stacked Deep LSTM

Complex hand gesture interactions among dynamic sign words may lead to misclassification, which affects the recognition accuracy of the ubiquitous sign language recognition system. This paper proposes to augment the feature vector of dynamic sign words with knowledge of hand dynamics as a proxy and classify dynamic sign words using motion patterns based on the extracted feature vector. In this method, some double-hand dynamic sign words have ambiguous or similar features across a hand motion trajectory, which leads to classification errors. Thus, the similar/ambiguous hand motion trajectory is determined based on the approximation of a probability density function over a time frame. Then, the extracted features are enhanced by transformation using maximal information correlation. These enhanced features of 3D skeletal videos captured by a leap motion controller are fed as a state transition pattern to a classifier for sign word classification. To evaluate the performance of the proposed method, an experiment is performed with 10 participants on 40 double hands dynamic ASL words, which reveals 97.98% accuracy. The method is further developed on challenging ASL, SHREC, and LMDHG data sets and outperforms conventional methods by 1.47%, 1.56%, and 0.37%, respectively.

FL-PMI: Federated Learning-Based Person Movement Identification through Wearable Devices in Smart Healthcare Systems

Recent technological developments, such as the Internet of Things (IoT), artificial intelligence, edge, and cloud computing, have paved the way in transforming traditional healthcare systems into smart healthcare (SHC) systems. SHC escalates healthcare management with increased efficiency, convenience, and personalization, via use of wearable devices and connectivity, to access information with rapid responses. Wearable devices are equipped with multiple sensors to identify a person’s movements. The unlabeled data acquired from these sensors are directly trained in the cloud servers, which require vast memory and high computational costs. To overcome this limitation in SHC, we propose a federated learning-based person movement identification (FL-PMI). The deep reinforcement learning (DRL) framework is leveraged in FL-PMI for auto-labeling the unlabeled data. The data are then trained using federated learning (FL), in which the edge servers allow the parameters alone to pass on the cloud, rather than passing vast amounts of sensor data. Finally, the bidirectional long short-term memory (BiLSTM) in FL-PMI classifies the data for various processes associated with the SHC. The simulation results proved the efficiency of FL-PMI, with 99.67% accuracy scores, minimized memory usage and computational costs, and reduced transmission data by 36.73%.

Vehicle Destination Prediction Using Bidirectional LSTM with Attention Mechanism

Satellite navigation has become ubiquitous to plan and track travelling. Having access to a vehicle’s position enables the prediction of its destination. This opens the possibility to various benefits, such as early warnings of potential hazards, route diversions to pass traffic congestion, and optimizing fuel consumption for hybrid vehicles. Thus, reliably predicting destinations can bring benefits to the transportation industry. This paper investigates using deep learning methods for predicting a vehicle’s destination based on its journey history. With this aim, Dense Neural Networks (DNNs), Long Short-Term Memory (LSTM) networks, Bidirectional LSTM (BiLSTM), and networks with and without attention mechanisms are tested. Especially, LSTM and BiLSTM models with attention mechanism are commonly used for natural language processing and text-classification-related applications. On the other hand, this paper demonstrates the viability of these techniques in the automotive and associated industrial domain, aimed at generating industrial impact. The results of using satellite navigation data show that the BiLSTM with an attention mechanism exhibits better prediction performance destination, achieving an average accuracy of 96% against the test set (4% higher than the average accuracy of the standard BiLSTM) and consistently outperforming the other models by maintaining robustness and stability during forecasting.

BiLSTM-5mC: A Bidirectional Long Short-Term Memory-Based Approach for Predicting 5-Methylcytosine Sites in Genome-Wide DNA Promoters

An important reason of cancer proliferation is the change in DNA methylation patterns, characterized by the localized hypermethylation of the promoters of tumor-suppressor genes together with an overall decrease in the level of 5-methylcytosine (5mC). Therefore, identifying the 5mC sites in the promoters is a critical step towards further understanding the diverse functions of DNA methylation in genetic diseases such as cancers and aging. However, most wet-lab experimental techniques are often time consuming and laborious for detecting 5mC sites. In this study, we proposed a deep learning-based approach, called BiLSTM-5mC, for accurately identifying 5mC sites in genome-wide DNA promoters. First, we randomly divided the negative samples into 11 subsets of equal size, one of which can form the balance subset by combining with the positive samples in the same amount. Then, two types of feature vectors encoded by the one-hot method, and the nucleotide property and frequency (NPF) methods were fed into a bidirectional long short-term memory (BiLSTM) network and a full connection layer to train the 22 submodels. Finally, the outputs of these models were integrated to predict 5mC sites by using the majority vote strategy. Our experimental results demonstrated that BiLSTM-5mC outperformed existing methods based on the same independent dataset.

Improved Prediction Model of Protein Lysine Crotonylation Sites Using Bidirectional Recurrent Neural Networks

Histone lysine crotonylation (Kcr) is a post-translational modification of histone proteins that is involved in the regulation of gene transcription, acute and chronic kidney injury, spermatogenesis, depression, cancer, and so forth. The identification of Kcr sites in proteins is important for characterizing and regulating primary biological mechanisms. The use of computational approaches such as machine learning and deep learning algorithms have emerged in recent years as the traditional wet-lab experiments are time-consuming and costly. We propose as part of this study a deep learning model based on a recurrent neural network (RNN) termed as Sohoko-Kcr for the prediction of Kcr sites. Through the embedded encoding of the peptide sequences, we investigate the efficiency of RNN-based models such as long short-term memory (LSTM), bidirectional LSTM (BiLSTM), and bidirectional gated recurrent unit (BiGRU) networks using cross-validation and independent tests. We also established the comparison between Sohoko-Kcr and other published tools to verify the efficiency of our model based on 3-fold, 5-fold, and 10-fold cross-validations using independent set tests. The results then show that the BiGRU model has consistently displayed outstanding performance and computational efficiency. Based on the proposed model, a webserver called Sohoko-Kcr was deployed for free use and is accessible at https://sohoko-research-9uu23.ondigitalocean.app.

Patient-Independent Seizure Detection Based on Channel-Perturbation Convolutional Neural Network and Bidirectional Long Short-Term Memory

Automatic seizure detection is of great significance for epilepsy diagnosis and alleviating the massive burden caused by manual inspection of long-term EEG. At present, most seizure detection methods are highly patient-dependent and have poor generalization performance. In this study, a novel patient-independent approach is proposed to effectively detect seizure onsets. First, the multi-channel EEG recordings are preprocessed by wavelet decomposition. Then, the Convolutional Neural Network (CNN) with proper depth works as an EEG feature extractor. Next, the obtained features are fed into a Bidirectional Long Short-Term Memory (BiLSTM) network to further capture the temporal variation characteristics. Finally, aiming to reduce the false detection rate (FDR) and improve the sensitivity, the postprocessing, including smoothing and collar, is performed on the outputs of the model. During the training stage, a novel channel perturbation technique is introduced to enhance the model generalization ability. The proposed approach is comprehensively evaluated on the CHB-MIT public scalp EEG database as well as a more challenging SH-SDU scalp EEG database we collected. Segment-based average accuracies of 97.51% and 93.70%, event-based average sensitivities of 86.51% and 89.89%, and average AUC-ROC of 90.82% and 90.75% are yielded on the CHB-MIT database and SH-SDU database, respectively.

prPred-DRLF: Plant R protein predictor using deep representation learning features

Plant resistance (R) proteins play a significant role in the detection of pathogen invasion. Accurately predicting plant R proteins is a key task in phytopathology. Most plant R protein predictors are dependent on traditional feature extraction methods. Recently, deep representation learning methods have been successfully applied in solving protein classification problems. Motivated by this, we propose a new computational approach, called prPred-DRLF, which uses deep representation learning feature models to encode the amino acids as numerical vectors. The results show that the fused features of bidirectional long short-term memory (BiLSTM) embedding and unified representation (UniRep) embedding have a better performance than other features for plant R protein identification using a light gradient boosting machine (LGBM) classifier. The model was evaluated using an independent test achieving an accuracy of 0.956, F1-score of 0.933, and area under the receiver operating characteristic (ROC) curve (AUC) of 0.997. Meanwhile, compared with the state-of-the-art prPred and HMMER method, prPred-DRLF shows an overall improvement in accuracy, F1-score, AUC, and recall. prPred-DRLF is a higher-performance plant R protein prediction tool based on two kinds of deep representation learning technologies and offers a user-friendly interface for inspecting possible plant R proteins. We hope that prPred-DRLF will become a useful tool for biological research. A user-friendly webserver for prPred-DRLF is freely accessible at http://lab.malab.cn/soft/prPred-DRLF. The Python script can be downloaded from https://github.com/Wangys-prog/prPred-DRLF.

Deep Learning in Automatic Sleep Staging With a Single Channel Electroencephalography

This study centers on automatic sleep staging with a single channel electroencephalography (EEG), with some significant findings for sleep staging. In this study, we proposed a deep learning-based network by integrating attention mechanism and bidirectional long short-term memory neural network (AT-BiLSTM) to classify wakefulness, rapid eye movement (REM) sleep and non-REM (NREM) sleep stages N1, N2 and N3. The AT-BiLSTM network outperformed five other networks and achieved an accuracy of 83.78%, a Cohen's kappa coefficient of 0.766 and a macro F1-score of 82.14% on the PhysioNet Sleep-EDF Expanded dataset, and an accuracy of 81.72%, a Cohen's kappa coefficient of 0.751 and a macro F1-score of 80.74% on the DREAMS Subjects dataset. The proposed AT-BiLSTM network even achieved a higher accuracy than the existing methods based on traditional feature extraction. Moreover, better performance was obtained by the AT-BiLSTM network with the frontal EEG derivations than with EEG channels located at the central, occipital or parietal lobe. As EEG signal can be easily acquired using dry electrodes on the forehead, our findings might provide a promising solution for automatic sleep scoring without feature extraction and may prove very useful for the screening of sleep disorders.

Classification of Biomedical Texts for Cardiovascular Diseases with Deep Neural Network Using a Weighted Feature Representation Method

This study aims to improve the performance of multiclass classification of biomedical texts for cardiovascular diseases by combining two different feature representation methods, i.e., bag-of-words (BoW) and word embeddings (WE). To hybridize the two feature representations, we investigated a set of possible statistical weighting schemes to combine with each element of WE vectors, which were term frequency (TF), inverse document frequency (IDF) and class probability (CP) methods. Thus, we built a multiclass classification model using a bidirectional long short-term memory (BLSTM) with deep neural networks for all investigated operations of feature vector combinations. We used MIMIC III and the PubMed dataset for the developing language model. To evaluate the performance of our weighted feature representation approaches, we conducted a set of experiments for examining multiclass classification performance with the deep neural network model and other state-of-the-art machine learning (ML) approaches. In all experiments, we used the OHSUMED-400 dataset, which includes PubMed abstracts related with specifically one class over 23 cardiovascular disease categories. Afterwards, we presented the results obtained from experiments and provided a comparison with related research in the literature. The results of the experiment showed that our BLSTM model with the weighting techniques outperformed the baseline and other machine learning approaches in terms of validation accuracy. Finally, our model outperformed the scores of related studies in the literature. This study shows that weighted feature representation improves the performance of the multiclass classification.

Remaining Useful Life Prediction of Airplane Engine Based on PCA-BLSTM

The accurate prediction of airplane engine failure can provide a reasonable decision basis for airplane engine maintenance, effectively reducing maintenance costs and reducing the incidence of failure. According to the characteristics of the monitoring data of airplane engine sensors, this work proposed a remaining useful life (RUL) prediction model based on principal component analysis and bidirectional long short-term memory. Principal component analysis is used for feature extraction to remove useless information and noise. After this, bidirectional long short-term memory is used to learn the relationship between the state monitoring data and remaining useful life. This work includes data preprocessing, the construction of a hybrid model, the use of the NASA's Commercial Aerodynamic System Simulation (C-MAPSS) data set for training and testing, and the comparison of results with those of support vector regression, long short-term memory and bidirectional long short-term memory models. The hybrid model shows better prediction accuracy and performance, which can provide a basis for formulating a reasonable airplane engine health management plan.

Silent Speech Decoding Using Spectrogram Features Based on Neuromuscular Activities

Silent speech decoding is a novel application of the Brain-Computer Interface (BCI) based on articulatory neuromuscular activities, reducing difficulties in data acquirement and processing. In this paper, spatial features and decoders that can be used to recognize the neuromuscular signals are investigated. Surface electromyography (sEMG) data are recorded from human subjects in mimed speech situations. Specifically, we propose to utilize transfer learning and deep learning methods by transforming the sEMG data into spectrograms that contain abundant information in time and frequency domains and are regarded as channel-interactive. For transfer learning, a pre-trained model of Xception on the large image dataset is used for feature generation. Three deep learning methods, Multi-Layer Perception, Convolutional Neural Network and bidirectional Long Short-Term Memory, are then trained using the extracted features and evaluated for recognizing the articulatory muscles' movements in our word set. The proposed decoders successfully recognized the silent speech and bidirectional Long Short-Term Memory achieved the best accuracy of 90%, outperforming the other two algorithms. Experimental results demonstrate the validity of spectrogram features and deep learning algorithms.

RFPR-IDP: reduce the false positive rates for intrinsically disordered protein and region prediction by incorporating both fully ordered proteins and disordered proteins

As an important type of proteins, intrinsically disordered proteins/regions (IDPs/IDRs) are related to many crucial biological functions. Accurate prediction of IDPs/IDRs is beneficial to the prediction of protein structures and functions. Most of the existing methods ignore the fully ordered proteins without IDRs during training and test processes. As a result, the corresponding predictors prefer to predict the fully ordered proteins as disordered proteins. Unfortunately, these methods were only evaluated on datasets consisting of disordered proteins without or with only a few fully ordered proteins, and therefore, this problem escapes the attention of the researchers. However, most of the newly sequenced proteins are fully ordered proteins in nature. These predictors fail to accurately predict the ordered and disordered proteins in real-world applications. In this regard, we propose a new method called RFPR-IDP trained with both fully ordered proteins and disordered proteins, which is constructed based on the combination of convolution neural network (CNN) and bidirectional long short-term memory (BiLSTM). The experimental results show that although the existing predictors perform well for predicting the disordered proteins, they tend to predict the fully ordered proteins as disordered proteins. In contrast, the RFPR-IDP predictor can correctly predict the fully ordered proteins and outperform the other 10 state-of-the-art methods when evaluated on a test dataset with both fully ordered proteins and disordered proteins. The web server and datasets of RFPR-IDP are freely available at http://bliulab.net/RFPR-IDP/server.

Convolutional Neural Network and Bidirectional Long Short-Term Memory-Based Method for Predicting Drug-Disease Associations

Identifying novel indications for approved drugs can accelerate drug development and reduce research costs. Most previous studies used shallow models for prioritizing the potential drug-related diseases and failed to deeply integrate the paths between drugs and diseases which may contain additional association information. A deep-learning-based method for predicting drug-disease associations by integrating useful information is needed. We proposed a novel method based on a convolutional neural network (CNN) and bidirectional long short-term memory (BiLSTM)-CBPred-for predicting drug-related diseases. Our method deeply integrates similarities and associations between drugs and diseases, and paths among drug-disease pairs. The CNN-based framework focuses on learning the original representation of a drug-disease pair from their similarities and associations. As the drug-disease association possibility also depends on the multiple paths between them, the BiLSTM-based framework mainly learns the path representation of the drug-disease pair. In addition, considering that different paths have discriminate contributions to the association prediction, an attention mechanism at path level is constructed. Our method, CBPred, showed better performance and retrieved more real associations in the front of the results, which is more important for biologists. Case studies further confirmed that CBPred can discover potential drug-disease associations.

DeepDSSR: Deep Learning Structure for Human Donor Splice Sites Recognition

Human genes often, through alternative splicing of pre-messenger RNAs, produce multiple mRNAs and protein isoforms that may have similar or completely different functions. Identification of splice sites is, therefore, crucial to understand the gene structure and variants of mRNA and protein isoforms produced by the primary RNA transcripts. Although many computational methods have been developed to detect the splice sites in humans, this is still substantially a challenging problem and further improvement of the computational model is still foreseeable. Accordingly, we developed DeepDSSR (deep donor splice site recognizer), a novel deep learning based architecture, for predicting human donor splice sites. The proposed method, built upon publicly available and highly imbalanced benchmark dataset, is comparable with the leading deep learning based methods for detecting human donor splice sites. Performance evaluation metrics show that DeepDSSR outperformed the existing deep learning based methods. Future work will improve the predictive capabilities of our model, and we will build a model for the prediction of acceptor splice sites.