Electroencephalogram (EEG)-based applications often require numerous channels to achieve high performance, which limits their widespread use. Various channel selection methods have been proposed to identify minimum EEG channels without compromising performance. However, most methods are limited to specific data paradigms or prediction models. We propose NeuroXAI, a novel method that identifies channel importance regardless of the type of EEG application. It integrates the surrogate analysis algorithm to optimize EEG signals and the data sampling algorithm, which effectively selects from highly voluminous EEG data. The efficacy of channel selection via the proposed method was evaluated through three datasets acquired under different paradigms (motor imagery, steady-state visually evoked potentials, and event-related potentials). On datasets based on these paradigms, NeuroXAI-based channel selection reduced the number of channels while maintaining or enhancing performance. The advantages of the proposed method include enhanced performance, robustness over varying data paradigms and the type of prediction model. The XAI technique enables intuitive interpretation of the constructed model operation, making it applicable in various fields such as model debugging and model interpretation. NeuroXAI has the potential to be used as a practical tool to develop better EEG applications.

Sleep is essential for maintaining human health and quality of life. Analyzing physiological signals during sleep is critical in assessing sleep quality and diagnosing sleep disorders. However, manual diagnoses by clinicians are time-intensive and subjective. Despite advances in deep learning that have enhanced automation, these approaches remain heavily dependent on large-scale labeled datasets. This study introduces SynthSleepNet, a multimodal hybrid-self-supervised learning (SSL) framework designed for analyzing polysomnography (PSG) data. SynthSleepNet effectively integrates masked prediction and contrastive learning to leverage complementary features across multiple modalities, including electroencephalogram (EEG), electrooculography (EOG), electromyography (EMG), and electrocardiogram (ECG). This approach enables the model to learn highly expressive representations of PSG data. Furthermore, a temporal context module based on Mamba was developed to efficiently capture contextual information across signals. SynthSleepNet achieved superior performance compared to state-of-the-art methods across three downstream tasks: sleep-stage classification, apnea detection, and hypopnea detection, with accuracies of 89.89%, 99.75%, and 89.60%, respectively. The model demonstrated robust performance in a semi-SSL environment with limited labels, achieving accuracies of 87.98%, 99.37%, and 77.52% in the same tasks. These results underscore the potential of the model as a foundational tool for the comprehensive analysis of PSG data. SynthSleepNet demonstrates comprehensively superior performance across multiple downstream tasks compared to other methodologies, making it expected to set a new standard for sleep disorder monitoring and diagnostic systems. The source code is available at https://github.com/dlcjfgmlnasa/SynthSleepNet.

In the neurological intensive care setting implementing tier-based therapy, osmotic therapy serves as an intermediate approach for reducing intracranial pressure (ICP). However, discussions regarding quantitative and specific criteria for the maintenance and transition of treatment remain limited. This study proposes an explainable neural network forecasting of physiological responses after osmotic therapy administration for the purpose of treatment guidance. Recordings of arterial blood pressure and intracranial pressure were obtained from 107 patients with traumatic brain injury. To maintain adequate intracranial pressure, arterial blood pressure, and cerebral perfusion pressure, 20 % mannitol (100 ml; no. infusion, 3,571) and 11.3 % hypertonic saline (40 ml; no. infusion, 574) were administered. The LTSF-NLinear model was used to predict the target signal response for the subsequent 1 h. Six prediction models were derived by combining two agents (mannitol or hypertonic saline) and three target signals (mean arterial blood pressure, mean intracranial pressure and cerebral perfusion pressure). To validate the model’s robustness while considering patient heterogeneity, group (5)-fold cross-validation was implemented. To assess the predictive ability of the proposed model for adverse clinical events, binary classification of hypotension, intracranial hypertension, and cerebral hypoperfusion was performed. The overall function of the model was elucidated using linear weight visualization. Additionally, the DeepSHAP application allows for a detailed investigation of the prediction process for a single case. The relationship between response to osmotic therapy and patient outcomes was statistically significant. The proposed models achieved reasonable performance, with R2 score > 0.8 for physiological signal predictions and accuracy > 0.9 for event predictions. Through global explanation, the recent values of mean arterial blood pressure, mean intracranial pressure, and cerebral perfusion pressure of input signals significantly influenced physiological responses after osmotic treatment. In the context of local explanation (a single case), it was possible to determine how physiological state influenced the response. The proposed model is expected to provide objective and quantitative information on osmotherapy in neurological intensive care environments, enabling guidance for alternative and aggressive treatments. Thus, it can potentially aid in enhancing the prognosis of patients with traumatic brain injury and improving the clinical workflow. Moreover, the global and local explanations of the proposed model can provide valuable insights for future researchers.

There is a strong association between intracranial hypertension (IH) that occurs following the acute phase of traumatic brain injury (TBI) and negative outcomes. This study proposes a pressure-time dose (PTD)-based parameter that may specify a possible serious IH (SIH) event and develops a model to predict SIH. The minute-by-minute signals of arterial blood pressure (ABP) and intracranial pressure (ICP) of 117 TBI patients were utilized as the internal validation dataset. The SIH event was explored through the prognostic power of the IH event variables for the outcome after 6 months, and an IH event with thresholds that included an ICP of 20 mmHg and PTD > 130 mmHg * minutes was considered an SIH event. The physiological characteristics of normal, IH and SIH events were investigated. LightGBM was employed to forecast an SIH event from various time intervals using physiological parameters derived from the ABP and ICP. Training and validation were conducted on 1921 SIH events. External validation was performed on two multi-center datasets containing 26 and 382 SIH events. The SIH parameters could be used to predict mortality (AUROC = 0.893, p < 0.001) and favorability (AUROC = 0.858, p < 0.001). The trained model robustly forecasted SIH after 5 and 480 minutes with an accuracy of 86.95% and 72.18% in internal validation. External validation also revealed a similar performance. This study demonstrated that the proposed SIH prediction model has reasonable predictive capacities. A future intervention study is required to investigate whether the definition of SIH is maintained in multi-center data and to ensure the effects of the predictive system on TBI patient outcomes at the bedside.

Arterial blood pressure (ABP) monitoring may permit the early diagnosis and management of cardiovascular disease (CVD); however, existing methods for measuring ABP outside the clinic use inconvenient cuff sphygmomanometry, or do not estimate continuous ABP waveforms. This study proposes a novel deep learning model DeepCNAP for estimating continuous BP waveform from a noninvasively measured photoplethysmography (PPG) signal in real-time. DeepCNAP was designed through the combination of deep convolutional networks and self-attention. The proposed method was constructed via 10-fold cross-validation based on the MIMIC database (the number of subjects = 942, recording time = 374.43 hours). The performance of DeepCNAP was evaluated from two perspectives: estimating ABP from PPG and classifying hemodynamically unstable events (i.e., hypertension, prehypertension, hypotension, and the normal state). The mean absolute errors of the BP estimates were 3.40 ± 4.36 mmHg for systolic BP, 1.75 ± 2.25 mmHg for diastolic BP, and 3.23 ± 2.21 mmHg for the BP waveform, indicating that DeepCNAP satisfies the standards of both the British Hypertension Society (BHS) and the Association for the Advancement of Medical Instrumentation (AAMI). From the estimated BP, hypertension, prehypertension, hypotension, and the normal state were classified with 99.44, 97.58, 92.23, and 94.64% accuracy, respectively. DeepCNAP enables feasible real-time estimation of invasively measured ABP from noninvasive PPG. With its noninvasive nature, high accuracy, and clinical relevance, the proposed model could be valuable in general wards at hospitals and for wearable devices in daily life.

Intracranial hypertension (IH) following acute phase traumatic brain injury (TBI) is associated with high mortality. Objective: This study proposes a novel parameter that may identify a potentially life-threatening IH (LTH) event and designs a machine learning model to predict LTH. Continuous recordings of intracranial pressure (ICP) and arterial blood pressure (ABP) from 273 TBI patients were used as the development dataset. The pressure-time dose (PTD) and pressure reactivity index (PRx) were calculated for each IH event, and an IH event with PRx > 0 and PTD > 5 was considered an LTH event. The association between the LTH parameters accumulated over five days and mortality was analyzed. A categorical boosting (CatBoost) model was employed to predict the occurrence of a future LTH event from the onset of IH using the ABP- and ICP-related parameters. Training and validation were performed on a total of 5,938 IH events. External performance evaluation was performed in 307 IH events included in the Cerebral Haemodynamic Autoregulatory Information System (CHARIS) database. The performance of the proposed model was evaluated through the area under the receiver operating characteristic curve (AUROC). The LTH parameters were able to distinguish between the deceased and surviving patients (AUROC > 0.7, p < 0.001). The CatBoost model predicted LTH with an AUROC = 0.7 on the external test dataset. This study demonstrated that the proposed LTH prediction model has a reasonable predictive capacity for mortality. The CatBoost model anticipates whether an IH event will develop into an LTH event. The findings of this study support the usefulness of ICP monitoring.

Motor imagery (MI) classification is important as the emerging research interest of brain computer interface (BCI) due to its potential about real-world application. Advancing manipulation and control technology of external devices such as robotics, the need of MI for complex and human-like movements is growing. The two most important procedures that influence the performance of MI-BCI are feature extraction and classification. Although there have been recent studies on feature extraction for complex, there is no consensus on the classifier suitable for complex MI. This study aimed to identify the best classifier for complex MI decoding.Electroencephalography (EEG) recordings measured during complex MI, which are hand grasping, spreading, pronation and supination, were used for binary (grasp vs. twist) and quaternary classification. Time domain parameter, which have shown suitability for complex movement decoding in previous works, was used as the EEG feature. Four types of ten machine learning classifiers, which have been applied to MI-BCI, were compared.Shrinkage regularized linear discriminant analysis (SRLDA) exhibited the best classification accuracy in both binary (92.8%) and quaternary (55.2%). In the case of training and testing time, a small amount of time for real-time analysis were needed, except random forest and logistic regression.This study showed that SRLDA is an appropriate classifier for complex MI classification, due to its ability to handle stationary and high dimensionality feature, TDP. The findings suggest that complex MI-BCI could gain more benefit from applying linear and shrinkage regularized model (i.e., SRLDA).

Electroencephalography (EEG) is a primary modality for estimating user intention in the brain-computer interface (BCI). In particular, BCI has been widely used to detect the intention of users in motor imagery (MI)-based tasks. Although the MI classification accuracy has been largely enhanced from previous efforts, MI-BCI studies have focused on extracting features only during MI tasks, not during the preparatory phases. The increment of alpha band power is induced by performing a task with attention. This study proposes a n approach for increasing MI-BCI performance by analyzing brain state in preparatory before the task. EEG recordings of nine healthy subjects from the open BCI dataset were investigated. The alpha lateralization index (ALI) was calculated for each trial and high ALI trials were utilized for learning lateralization-based model. MI classification accuracy using the lateralization-based model marked high performance (median accuracy = 63.2 %; interquartile range (IQR) = 50.0% - 54.8%) than the total trial-based approach (median accuracy = 52.0%; IQR = 50.0 % - 54.8%) with statistical significance (p = 0.018). This study suggests alpha lateralization which is an imbalance pattern between ipsilateral and contralateral is one of the main factors for improvement of performance. Accordingly, since the alpha liberalization before MI task could exert an effect on the MI phase, the analysis combined preparation with MI would derive highly benefit for the MI classification.

Electroencephalography (EEG) is the primary modality for estimating the user intention in brain-computer interface (BCI). However, the suppression of the inter-subject variability (ISV) remains as a major challenge in constructing a reliable EEG-based BCI model. Subject-specific classification models have been widely used to avoid ISV, however these inherently involve time-consuming individual calibration process. This study speculated that the calibration could be minimized via clustering BCI subjects into subgroups by their respective similarity in brain power distribution at the resting state and conducted a proof-of-concept investigation. EEG recordings of twenty-nine healthy subjects from open motor imagery (MI) dataset were used in this study. K-means clustering based on brain activation in α, low β and high β-band at resting state divided the subjects into three subgroups. The efficacy of band-clustering was evaluated by comparing its MI classification performance (leftor right-hand gripping) to subject-specific and general models. Among the subjects in a cluster, ISV was lower than that in twenty-nine subjects, especially in the α-band. The MI classification accuracy using the cluster-specific model on the α-band was marked high performance (median accuracy 68.8%). The cluster-specific model had significantly high accuracy compared to general model (median accuracy = 64.6%). Furthermore, the difference of MI classification accuracy between the cluster-specific model on the α band and subject-specific model is not significant (median accuracy = 69.3%). Consequently, establishing a model by grouping clusters using similar brain activation patterns was highly beneficial for the MI classification without individual calibration process.

Near-infrared spectroscopy (NIRS) is receiving much attention in the fields of brain-computer interface (BCI) due to its noninvasiveness, usability, and high performance. However, NIRS is susceptible to the motion artifacts, and its high morphological variety hinders the existing signals artifacts elimination techniques from being applied to NIRS. This study proposes a novel feature extraction method for classifying motion artifacts. NIRS data from an open access dataset containing five types of motion artifact (i.e., eye blinking, head movement, eyes movement, teeth clenching, and mouth opening) from 28 healthy subjects were analyzed. The efficacy of the proposed wavelet statistical feature extraction method in artifact classification was compared to various existing feature extraction methods for BCI designs. Each feature was learned by four conventional machine learning models for classifying NIRS motion artifacts. Shrinkage regularized linear discriminant analysis (SRLDA) with the proposed features derived from oxy and deoxy-hemoglobin NIRS signal achieved 90% accuracy and 0.85 Cohen's kappa coefficient for classifying types of motion artifacts. Mann-Whitney U test and paired T-test indicate that SRLDA with the proposed feature had significantly higher classification performance compared to the other models. The proposed method can reliably classify the motion artifacts to standardize the morphology types of NIRS artifact. It could be used to augment signal quality control technique with further extension in future NIRS-based BCI systems.

The development of brain-computer interface (BCI) systems that are based on electroencephalography (EEG), and driven by spontaneous movement intentions, is useful for rehabilitation and external device control. In this study, we analyzed the decoding of five different hand executions and imageries from EEG signals, for a robot hand control. Five healthy subjects participated in this experiment. They executed and imagined five sustained hand motions. In this motor execution (ME) and motor imagery (MI) experiment, we proposed a subject-specific time interval selection method, and we used common spatial patterns (CSP) and the regularized linear discriminant analysis (RLDA) for the data analysis. As a result, we classified the five different hand motions offline and obtained average classification accuracies of 56.83% for ME, and 51.01% for MI, respectively. Both results were higher than the obtained accuracies from a comparison method that used a standard fixed time interval method. This result is encouraging, and the proposed method could potentially be used in future applications, such as a BCI-driven robot hand control.

This paper presents the feasibility of an electroencephalography (EEG)-based robot arm control system using a decoding of multi-directional arm reaching movement imagery. To do that, we have designed and implemented an experimental environment that can acquire non-invasive brain signals about multi-directional arm reaching movement. Five subjects participated in our experiments and the subjects performed four directional reaching tasks (Left, right, forward, and backward) with actual movement and movement imagery. The filter-bank common spatial pattern (FBCSP) was applied to extract spatio-frequency features from the acquired EEG signals. The regularized linear discriminant analysis (RLDA) was also applied as a classifier. As a result, the averaged classification accuracies of the actual movement and movement imagery were represented 67.04% and 59.19%, respectively. These results showed a feasibility of the EEG-based robot arm control system based on multi-directional arm reaching movement imagery.

In this paper, we propose a simple frequency-domain interpolation technique for the discrete Fourier transform (DFT). This interpolation technique can significantly improve the frequency and phase resolution capabilities of the DFT without increasing the size of the DFT (the number of points used for the DFT). This new technique employs a dividing point in the amplitude and phase spectra. Suitable areas of application include joint estimation of fine frequency and phase offsets in burst-mode digital transmission.