Unsupervised Spiking Neural Network with Dynamic Learning of Inhibitory Neurons.

A spiking neural network (SNN) is a type of artificial neural network that operates based on discrete spikes to process timing information, similar to the manner in which the human brain processes real-world problems. In this paper, we propose a new spiking neural network (SNN) based on conventional, biologically plausible paradigms, such as the leaky integrate-and-fire model, spike timing-dependent plasticity, and the adaptive spiking threshold, by suggesting new biological models; that is, dynamic inhibition weight change, a synaptic wiring method, and Bayesian inference. The proposed network is designed for image recognition tasks, which are frequently used to evaluate the performance of conventional deep neural networks. To manifest the bio-realistic neural architecture, the learning is unsupervised, and the inhibition weight is dynamically changed; this, in turn, affects the synaptic wiring method based on Hebbian learning and the neuronal population. In the inference phase, Bayesian inference successfully classifies the input digits by counting the spikes from the responding neurons. The experimental results demonstrate that the proposed biological model ensures a performance improvement compared with other biologically plausible SNN models.

Blood pressure estimation and its recalibration assessment using wrist cuff blood pressure monitor.

The rapid evolution of wearable technology in healthcare sectors has created the opportunity for people to measure their blood pressure (BP) using a smartwatch at any time during their daily activities. Several commercially-available wearable devices have recently been equipped with a BP monitoring feature. However, concerns about recalibration remain. Pulse transit time (PTT)-based estimation is required for initial calibration, followed by periodic recalibration. Recalibration using arm-cuff BP monitors is not practical during everyday activities. In this study, we investigated recalibration using PTT-based BP monitoring aided by a deep neural network (DNN) and validated the performance achieved with more practical wrist-cuff BP monitors. The PTT-based prediction produced a mean absolute error (MAE) of 4.746 ± 1.529 mmHg for systolic blood pressure (SBP) and 3.448 ± 0.608 mmHg for diastolic blood pressure (DBP) when tested with an arm-cuff monitor employing recalibration. Recalibration clearly improved the performance of both DNN and conventional linear regression approaches. We established that the periodic recalibration performed by a wrist-worn BP monitor could be as accurate as that obtained with an arm-worn monitor, confirming the suitability of wrist-worn devices for everyday use. This is the first study to establish the potential of wrist-cuff BP monitors as a means to calibrate BP monitoring devices that can reliably substitute for arm-cuff BP monitors. With the use of wrist-cuff BP monitoring devices, continuous BP estimation, as well as frequent calibrations to ensure accurate BP monitoring, are now feasible.

GAN-based patient information hiding for an ECG authentication system.

Various biometrics such as the face, irises, and fingerprints, which can be obtained in a relatively simple way in modern society, are used in personal authentication systems to identify individuals. These biometric data are extracted from an individual's physiological data and yield high performance in identifying an individual using unique data patterns. Biometric identification is also used in portable devices such as mobile devices because it is more secure than cryptographic token-based authentication methods. However, physiological data could include personal health information such as arrhythmia related patterns in electrocardiogram (ECG) signals. To protect sensitive health information from hackers, the biomarkers of certain diseases or disorders that exist in ECG signals need to be hidden. Additionally, to implement the inference models for both arrhythmia detection and personal authentication in a mobile device, a lightweight model such as a multi-task deep learning model should be considered. This study demonstrates a multi-task neural network model that simultaneously identifies an individual's ECG and arrhythmia patterns using a small network. Finally, the computational efficiency and model size of the single-task and multi-task models were compared based on the number of parameters. Although the multi-task model has 20,000 fewer parameters than the single-task model, they yielded similar performance, which demonstrates the efficient structure of the multi-task model.

Intelligent Feature Selection for ECG-Based Personal Authentication Using Deep Reinforcement Learning.

In this study, the optimal features of electrocardiogram (ECG) signals were investigated for the implementation of a personal authentication system using a reinforcement learning (RL) algorithm. ECG signals were recorded from 11 subjects for 6 days. Consecutive 5-day datasets (from the 1st to the 5th day) were trained, and the 6th dataset was tested. To search for the optimal features of ECG for the authentication problem, RL was utilized as an optimizer, and its internal model was designed based on deep learning structures. In addition, the deep learning architecture in RL was automatically constructed based on an optimization approach called Bayesian optimization hyperband. The experimental results demonstrate that the feature selection process is essential to improve the authentication performance with fewer features to implement an efficient system in terms of computation power and energy consumption for a wearable device intended to be used as an authentication system. Support vector machines in conjunction with the optimized RL algorithm yielded accuracy outcomes using fewer features that were approximately 5%, 3.6%, and 2.6% higher than those associated with information gain (IG), ReliefF, and pure reinforcement learning structures, respectively. Additionally, the optimized RL yielded mostly lower equal error rate (EER) values than the other feature selection algorithms, with fewer selected features.

Multitask Siamese Network for Remote Photoplethysmography and Respiration Estimation.

Heart and respiration rates represent important vital signs for the assessment of a person's health condition. To estimate these vital signs accurately, we propose a multitask Siamese network model (MTS) that combines the advantages of the Siamese network and the multitask learning architecture. The MTS model was trained by the images of the cheek including nose and mouth and forehead areas while sharing the same parameters between the Siamese networks, in order to extract the features about the heart and respiratory information. The proposed model was constructed with a small number of parameters and was able to yield a high vital-sign-prediction accuracy, comparable to that obtained from the single-task learning model; furthermore, the proposed model outperformed the conventional multitask learning model. As a result, we can simultaneously predict the heart and respiratory signals with the MTS model, while the number of parameters was reduced by 16 times with the mean average errors of heart and respiration rates being 2.84 and 4.21. Owing to its light weight, it would be advantageous to implement the vital-sign-monitoring model in an edge device such as a mobile phone or small-sized portable devices.

Effect of a Recliner Chair with Rocking Motions on Sleep Efficiency.

In this study, we analyze the effect of a recliner chair with rocking motions on sleep quality of naps using automated sleep scoring and spindle detection models. The quality of sleep corresponding to the two rocking motions was measured quantitatively and qualitatively. For the quantitative evaluation, we conducted a sleep parameter analysis based on the results of the estimated sleep stages obtained on the brainwave and spindle estimation, and a sleep survey assessment from the participants was analyzed for the qualitative evaluation. The analysis showed that sleep in the recliner chair with rocking motions positively increased the duration of the spindles and deep sleep stage, resulting in improved sleep quality.

Deep Learning-Based Optimal Smart Shoes Sensor Selection for Energy Expenditure and Heart Rate Estimation.

Wearable technologies are known to improve our quality of life. Among the various wearable devices, shoes are non-intrusive, lightweight, and can be used for outdoor activities. In this study, we estimated the energy consumption and heart rate in an environment (i.e., running on a treadmill) using smart shoes equipped with triaxial acceleration, triaxial gyroscope, and four-point pressure sensors. The proposed model uses the latest deep learning architecture which does not require any separate preprocessing. Moreover, it is possible to select the optimal sensor using a channel-wise attention mechanism to weigh the sensors depending on their contributions to the estimation of energy expenditure (EE) and heart rate (HR). The performance of the proposed model was evaluated using the root mean squared error (RMSE), mean absolute error (MAE), and coefficient of determination (R2). Moreover, the RMSE was 1.05 ± 0.15, MAE 0.83 ± 0.12 and R2 0.922 ± 0.005 in EE estimation. On the other hand, and RMSE was 7.87 ± 1.12, MAE 6.21 ± 0.86, and R2 0.897 ± 0.017 in HR estimation. In both estimations, the most effective sensor was the z axis of the accelerometer and gyroscope sensors. Through these results, it is demonstrated that the proposed model could contribute to the improvement of the performance of both EE and HR estimations by effectively selecting the optimal sensors during the active movements of participants.

Efficiently Updating ECG-Based Biometric Authentication Based on Incremental Learning.

Recently, the interest in biometric authentication based on electrocardiograms (ECGs) has increased. Nevertheless, the ECG signal of a person may vary according to factors such as the emotional or physical state, thus hindering authentication. We propose an adaptive ECG-based authentication method that performs incremental learning to identify ECG signals from a subject under a variety of measurement conditions. An incremental support vector machine (SVM) is adopted for authentication implementing incremental learning. We collected ECG signals from 11 subjects during 10 min over six days and used the data from days 1 to 5 for incremental learning, and those from day 6 for testing. The authentication results show that the proposed system consistently reduces the false acceptance rate from 6.49% to 4.39% and increases the true acceptance rate from 61.32% to 87.61% per single ECG wave after incremental learning using data from the five days. In addition, the authentication results tested using data obtained a day after the latest training show the false acceptance rate being within reliable range (3.5-5.33%) and improvement of the true acceptance rate (70.05-87.61%) over five days.

Beat-to-Beat Continuous Blood Pressure Estimation Using Bidirectional Long Short-Term Memory Network.

Continuous blood pressure (BP) monitoring is important for patients with hypertension. However, BP measurement with a cuff may be cumbersome for the patient. To overcome this limitation, various studies have suggested cuffless BP estimation models using deep learning algorithms. A generalized model should be considered to decrease the training time, and the model reproducibility should be taken into account in multi-day scenarios. In this study, a BP estimation model with a bidirectional long short-term memory network is proposed. The features are extracted from the electrocardiogram, photoplethysmogram, and ballistocardiogram. The leave-one-subject-out (LOSO) method is incorporated to generalize the model and fine-tuning is applied. The model was evaluated using one-day and multi-day tests. The proposed model achieved a mean absolute error (MAE) of 2.56 and 2.05 mmHg for the systolic and diastolic BP (SBP and DBP), respectively, in the one-day test. Moreover, the results demonstrated that the LOSO method with fine-tuning was more compatible in the multi-day test. The MAE values of the model were 5.82 and 5.24 mmHg for the SBP and DBP, respectively.

Automated White Blood Cell Counting in Nailfold Capillary Using Deep Learning Segmentation and Video Stabilization.

White blood cells (WBCs) are essential components of the immune system in the human body. Various invasive and noninvasive methods to monitor the condition of the WBCs have been developed. Among them, a noninvasive method exploits an optical characteristic of WBCs in a nailfold capillary image, as they appear as visual gaps. This method is inexpensive and could possibly be implemented on a portable device. However, recent studies on this method use a manual or semimanual image segmentation, which depends on recognizable features and the intervention of experts, hindering its scalability and applicability. We address and solve this problem with proposing an automated method for detecting and counting WBCs that appear as visual gaps on nailfold capillary images. The proposed method consists of an automatic capillary segmentation method using deep learning, video stabilization, and WBC event detection algorithms. Performances of the three segmentation algorithms (manual, conventional, and deep learning) with/without video stabilization were benchmarks. Experimental results demonstrate that the proposed method improves the performance of the WBC event counting and outperforms conventional approaches.

Feature Analysis of Smart Shoe Sensors for Classification of Gait Patterns.

Gait analysis is commonly used to detect foot disorders and abnormalities such as supination, pronation, unstable left foot and unstable right foot. Early detection of these abnormalities could help us to correct the walking posture and avoid getting injuries. This paper presents extensive feature analyses on smart shoes sensor data, including pressure sensors, accelerometer and gyroscope signals, to obtain the optimum combination of the sensors for gait classification, which is crucial to implement a power-efficient mobile smart shoes system. In addition, we investigated the optimal length of data segmentation based on the gait cycle parameters, reduction of the feature dimensions and feature selection for the classification of the gait patterns. Benchmark tests among several machine learning algorithms were conducted using random forest, k-nearest neighbor (KNN), logistic regression and support vector machine (SVM) algorithms for the classification task. Our experiments demonstrated the combination of accelerometer and gyroscope sensor features with SVM achieved the best performance with 89.36% accuracy, 89.76% precision and 88.44% recall. This research suggests a new state-of-the-art gait classification approach, specifically on detecting human gait abnormalities.

Effect of an Inflatable Air Mattress with Variable Rigidity on Sleep Quality.

Several studies, wherein the structure or rigidity of a mattress was varied, have been conducted to improve sleep quality. These studies investigated the effect of variation in the surface characteristics of mattresses on sleep quality. The present study developed a mattress whose rigidity can be varied by controlling the amount of air in its air cells. To investigate the effect of the variable rigidity of the air mattress on sleep quality, participants (Male, Age: 23.9 ± 2.74, BMI: 23.3 ± 1.60) were instructed to sleep on the air mattress under different conditions, and their sleep quality was subjectively and objectively investigated. Subjectively, sleep quality is assessed based on the participants' evaluations of the depth and length of their sleep. Objectively, sleep is estimated using the sleep stage information obtained by analysing the movements and brain waves of the participants during their sleep. A subjective assessment of the sleep quality demonstrates that the participants' sleep was worse with the adjustment of the air mattress than that without; however, the objective sleep quality results demonstrates an improvement in the sleep quality when the rigidity of the air mattress is varied based on the participant's preference. This paper proposes a design for mattresses that can result in more efficient sleep than that provided by traditional mattresses.

End-to-End Deep Learning Architecture for Continuous Blood Pressure Estimation Using Attention Mechanism.

Blood pressure (BP) is a vital sign that provides fundamental health information regarding patients. Continuous BP monitoring is important for patients with hypertension. Various studies have proposed cuff-less BP monitoring methods using pulse transit time. We propose an end-to-end deep learning architecture using only raw signals without the process of extracting features to improve the BP estimation performance using the attention mechanism. The proposed model consisted of a convolutional neural network, a bidirectional gated recurrent unit, and an attention mechanism. The model was trained by a calibration-based method, using the data of each subject. The performance of the model was compared to the model that used each combination of the three signals, and the model with the attention mechanism showed better performance than other state-of-the-art methods, including conventional linear regression method using pulse transit time (PTT). A total of 15 subjects were recruited, and electrocardiogram, ballistocardiogram, and photoplethysmogram levels were measured. The 95% confidence interval of the reference BP was [86.34, 143.74] and [51.28, 88.74] for systolic BP (SBP) and diastolic BP (DBP), respectively. The R 2 values were 0.52 and 0.49, and the mean-absolute-error values were 4.06 ± 4.04 and 3.33 ± 3.42 for SBP and DBP, respectively. In addition, the results complied with global standards. The results show the applicability of the proposed model as an analytical metric for BP estimation.

Correlation Assisted Strong Uncorrelating Transform Complex Common Spatial Patterns for Spatially Distant Channel Data.

The Strong Uncorrelating Transform Complex Common Spatial Patterns (SUTCCSP) algorithm, designed for multichannel data analysis, has a limitation on keeping the correlation information among channels during the simultaneous diagonalization process of the covariance and pseudocovariance matrices. This paper focuses on the importance of preserving the correlation information among multichannel data and proposes the correlation assisted SUTCCSP (CASUT) algorithm to address this issue. The performance of the proposed algorithm was demonstrated by classifying the motor imagery electroencephalogram (EEG) dataset. The features were first extracted using CSP algorithms including the proposed method, and then the random forest classifier was utilized for the classification. Experiments using CASUT yielded an average classification accuracy of 78.10 (%), which significantly outperformed those of original CSP, Complex Common Spatial Patterns (CCSP), and SUTCCSP with p-values less than 0.01, tested by the Wilcoxon signed rank test.

Deep ECGNet: An Optimal Deep Learning Framework for Monitoring Mental Stress Using Ultra Short-Term ECG Signals.

BACKGROUND: Stress recognition using electrocardiogram (ECG) signals requires the intractable long-term heart rate variability (HRV) parameter extraction process. This study proposes a novel deep learning framework to recognize the stressful states, the Deep ECGNet, using ultra short-term raw ECG signals without any feature engineering methods. METHODS: The Deep ECGNet was developed through various experiments and analysis of ECG waveforms. We proposed the optimal recurrent and convolutional neural networks architecture, and also the optimal convolution filter length (related to the P, Q, R, S, and T wave durations of ECG) and pooling length (related to the heart beat period) based on the optimization experiments and analysis on the waveform characteristics of ECG signals. The experiments were also conducted with conventional methods using HRV parameters and frequency features as a benchmark test. The data used in this study were obtained from Kwangwoon University in Korea (13 subjects, Case 1) and KU Leuven University in Belgium (9 subjects, Case 2). Experiments were designed according to various experimental protocols to elicit stressful conditions. RESULTS: The proposed framework to recognize stress conditions, the Deep ECGNet, outperformed the conventional approaches with the highest accuracy of 87.39% for Case 1 and 73.96% for Case 2, respectively, that is, 16.22% and 10.98% improvements compared with those of the conventional HRV method. CONCLUSIONS: We proposed an optimal deep learning architecture and its parameters for stress recognition, and the theoretical consideration on how to design the deep learning structure based on the periodic patterns of the raw ECG data. Experimental results in this study have proved that the proposed deep learning model, the Deep ECGNet, is an optimal structure to recognize the stress conditions using ultra short-term ECG data.

Motor Imagery Classification Using Mu and Beta Rhythms of EEG with Strong Uncorrelating Transform Based Complex Common Spatial Patterns.

Recent studies have demonstrated the disassociation between the mu and beta rhythms of electroencephalogram (EEG) during motor imagery tasks. The proposed algorithm in this paper uses a fully data-driven multivariate empirical mode decomposition (MEMD) in order to obtain the mu and beta rhythms from the nonlinear EEG signals. Then, the strong uncorrelating transform complex common spatial patterns (SUTCCSP) algorithm is applied to the rhythms so that the complex data, constructed with the mu and beta rhythms, becomes uncorrelated and its pseudocovariance provides supplementary power difference information between the two rhythms. The extracted features using SUTCCSP that maximize the interclass variances are classified using various classification algorithms for the separation of the left- and right-hand motor imagery EEG acquired from the Physionet database. This paper shows that the supplementary information of the power difference between mu and beta rhythms obtained using SUTCCSP provides an important feature for the classification of the left- and right-hand motor imagery tasks. In addition, MEMD is proved to be a preferred preprocessing method for the nonlinear and nonstationary EEG signals compared to the conventional IIR filtering. Finally, the random forest classifier yielded a high performance for the classification of the motor imagery tasks.

Self-Adhesive and Capacitive Carbon Nanotube-Based Electrode to Record Electroencephalograph Signals From the Hairy Scalp.

We fabricated a carbon nanotube (CNT)/adhesive polydimethylsiloxane (aPDMS) composite-based dry electroencephalograph (EEG) electrode for capacitive measuring of EEG signals. As research related to brain-computer interface applications has advanced, the presence of hairs on a patient's scalp has continued to present an obstacle to recorder EEG signals using dry electrodes. The CNT/aPDMS electrode developed here is elastic, highly conductive, self-adhesive, and capable of making conformal contact with and attaching to a hairy scalp. Onto the conductive disk, hundreds of conductive pillars coated with Parylene C insulation layer were fabricated. A CNT/aPDMS layer was attached on the disk to transmit biosignals to the pillar. The top of disk was designed to be solderable, which enables the electrode to connect with a variety of commercial EEG acquisition systems. The mechanical and electrical characteristics of the electrode were tested, and the performances of the electrodes were evaluated by recording EEGs, including alpha rhythms, auditory-evoked potentials, and steady-state visually-evoked potentials. The results revealed that the electrode provided a high signal-to-noise ratio with good tolerance for motion. Almost no leakage current was observed. Although preamplifiers with ultrahigh input impedance have been essential for previous capacitive electrodes, the EEGs were recorded here by directly connecting a commercially available EEG acquisition system to the electrode to yield high-quality signals comparable to those obtained using conventional wet electrodes.

EEG gamma band oscillations differentiate the planning of spatially directed movements of the arm versus eye: multivariate empirical mode decomposition analysis.

The neural dynamics underlying the coordination of spatially-directed limb and eye movements in humans is not well understood. Part of the difficulty has been a lack of signal processing tools suitable for the analysis of nonstationary electroencephalographic (EEG) signals. Here, we use multivariate empirical mode decomposition (MEMD), a data-driven approach that does not employ predefined basis functions. High-density EEG, and arm and eye movements were synchronously recorded in 10 subjects performing time-constrained reaching and/or eye movements. Subjects were allowed to move both the hand and the eyes, only the hand, or only the eyes following a 500-700 ms delay interval where the hand and gaze remained on a central fixation cross. An additional condition involved a nonspatially-directed "lift" movement of the hand. The neural activity during a 500 ms delay interval was decomposed into intrinsic mode functions (IMFs) using MEMD. Classification analysis revealed that gamma band (30 Hz) IMFs produced more classifiable features differentiating the EEG according to the different upcoming movements. A benchmark test using conventional algorithms demonstrated that MEMD was the best algorithm for extracting oscillatory bands from EEG, yielding the best classification of the different movement conditions. The gamma rhythm decomposed using MEMD showed a higher correlation with the eventual movement accuracy than any other band rhythm and than any other algorithm.

Augmented Complex Common Spatial Patterns for Classification of Noncircular EEG From Motor Imagery Tasks.

A novel augmented complex-valued common spatial pattern (CSP) algorithm is introduced in order to cater for general complex signals with noncircular probability distributions. This is a typical case in multichannel electroencephalogram (EEG), due to the power difference or correlation between the data channels, yet current methods only cater for a very restrictive class of circular data. The proposed complex-valued CSP algorithms account for the generality of complex noncircular data, by virtue of the use of augmented complex statistics and the strong-uncorrelating transform (SUT). Depending on the degree of power difference of complex signals, the analysis and simulations show that the SUT based algorithm maximizes the inter-class difference between two motor imagery tasks. Simulations on both synthetic noncircular sources and motor imagery experiments using real-world EEG support the approach.