People with autism spectrum disorder (ASD) often have cognitive impairments. Effective connectivity between different areas of the brain is essential for normal cognition. Electroencephalography (EEG) has been widely used in the detection of neurological diseases. Previous studies on detecting ASD with EEG data have focused on frequency-related features. Most of these studies have augmented data by splitting the dataset into time slices or sliding windows. However, such approaches to data augmentation may cause the testing data to be contaminated by the training data. To solve this problem, this study developed a novel method for detecting ASD with EEG data. This study quantified the functional connectivity of the subject's brain from EEG signals and defined the individual to be the unit of analysis. Publicly available EEG data were gathered from 97 and 92 subjects with ASD and typical development (TD), respectively, while they were at rest or performing a task. Time-series maps of brain functional connectivity were constructed, and the data were augmented using a deep convolutional generative adversarial network. In addition, a combined network for ASD detection, based on convolutional neural network (CNN) and long short-term memory (LSTM), was designed and implemented. Based on functional connectivity, the network achieved classification accuracies of 81.08% and 74.55% on resting state and task state data, respectively. In addition, we found that the functional connectivity of ASD differed from TD primarily in the short-distance functional connectivity of the parietal and occipital lobes and in the distant connections from the right temporoparietal junction region to the left posterior temporal lobe. This paper provides a new perspective for better utilizing EEG to understand ASD. The method proposed in our study is expected to be a reliable tool to assist in the diagnosis of ASD.

The human–machine interface (HMI) detects electrophysiological signals from the subject and controls the machine based on the signal information. However, most applications are still only in the testing stage and are generally unavailable to the public. In recent years, researchers have been devoted to making wearable HMI devices smarter and more comfortable. In this study, a wearable, intelligent eight-channel electromyography (EMG) signal–based system was designed to recognize 21 types of gestures. An analog front end (AFE) integrated chip (IC) was developed to detect the EMG signals, and an integrated EMG signal acquisition device integrating an elastic armband was fabricated. An SIAT database of 21 gestures was established by collecting EMG gesture signals from 10 volunteers. A lightweight 1D CNN model was constructed and subjected to individualized training by using the SIAT database. The maximum signal recognition accuracy was 89.96%, and the average model training time was 14 min 13 s. Given its small size, the model can be applied on lower-performance edge computing devices and is expected to be applied to smartphone terminals in the future. The source code is available at https://github.com/Siat-F9/EMG-Tools .

Cardiovascular disease (CVD) is a widespread disease and the leading cause of death worldwide. Home care is essential for patients with CVD, and it involves the daily monitoring of important CVD-related vital signs using methods including electrocardiography (ECG), heart rate monitoring, pulse oximetry (SpO2), and continuous blood pressure measurement. However, a wearable device that can monitor these parameters simultaneously remains unavailable; herein, we propose a lightweight, highly integrated sensor that can do so. In this sensor, an analog front end (AFE) integrated chip (IC) is implemented in the sensor to detect one-lead ECG and two-wavelength photoplethysmography (PPG) signals. The highly integrated IC minimizes both the size and power requirement of the sensor. Moreover, its comprehensive functions include adjustable gain, current compensation, lead off, and fast recovery—all of which are crucial for wearable applications in large populations. Accordingly, the sensor can apply the adaptive adjust method to automatically adjust the IC parameters to suit a range of applications and users. In addition, the heart rate is calculated from the R-R interval from the ECG signals, whereas the SpO2 is calibrated with a univariate quadratic equation with less than 1% mean error. Our sensor can also calculate the systolic blood pressure (SBP) and diastolic blood pressure (DBP) by using a support vector machine based calibration-less model with the features of infrared PPG and ECG signals. The model is trained on our pre-collected wearable dataset and has a mean error ± standard deviation of −2.10 ± 7.07 mmHg for SBP and 0.04 ± 7.34 mmHg for DBP in 16 volunteers. In conclusion, this paper reports on a multimodal, vital sign monitoring sensor—with small size, low power, and dynamic compatibility—suitable for patients with CVD under home care.

Hybrid motor imagery brain-computer interfaces (MI-BCIs), which integrate both electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) signals, outperform those based solely on EEG. However, simultaneously recording EEG and fNIRS signals is highly challenging due to the difficulty of colocating both types of sensors on the same scalp surface. This physical constraint complicates the acquisition of high-quality hybrid signals, thereby limiting the widespread application of hybrid MI-BCIs. To address this issue, this study proposes the spatio-temporal controlled diffusion model (SCDM) as a framework for cross-modal generation from EEG to fNIRS. The model utilizes two core modules, the spatial cross-modal generation (SCG) module and the multi-scale temporal representation (MTR) module, which adaptively learn the respective latent temporal and spatial representations of both signals in a unified representation space. The SCG module further maps EEG representations to fNIRS representations by leveraging their spatial relationships. Experimental results show high similarity between synthetic and real fNIRS signals. The joint classification performance of EEG and synthetic fNIRS signals is comparable to or even better than that of EEG with real fNIRS signals. Furthermore, the synthetic signals exhibit similar spatio-temporal features to real signals while preserving spatial relationships with EEG signals. To our knowledge, it is the first work that an end-to-end framework is proposed to achieve cross-modal generation from EEG to fNIRS. Experimental results suggest that the SCDM may represent a promising paradigm for the acquisition of hybrid EEG-fNIRS signals in MI-BCI systems.

In consumer electronics, affective computing allows devices to recognize and respond to the emotional states of individuals, thereby enhancing personalization and satisfaction. Electroencephalogram (EEG)-based emotion recognition methods have garnered attention for their capability to provide real-time, objective insights into emotional and cognitive states. Existing research mainly focuses on extracting temporal and spectral features from EEG signals for emotion recognition. However, electrode shifts may occur across different experiments during data acquisition, resulting in models with limited robustness and constrained classification performance. To address this challenge, this paper introduces a Brain Region Knowledge based Dual-Stream network (BRKDSnet) for emotion recognition. It employs a dual-stream architecture to effectively fuse the temporal, spectral, and spatial features of EEG signals. By incorporating prior brain region partition knowledge, it efficiently aggregates channel features into distinct brain region features. Such a brain region based partition mitigates noise issues caused by electrode shifts during user wearing in consumer electronic applications, and leading to learn more robust feature representations. The experiments conducted on the widely used emotion recognition datasets SEED and SEED-IV shows that BRKDSnet achieves state-of-the-art results, with accuracies of 97.81% (improved by 0.64%-3.57% compared to existing works) and 90.76% (improved by 3.13%-11.39%) respectively. Finally, through analysis of the visualization results, it is found that EEG channels located in the temporal lobe contributed the most to emotion recognition classification, which also provides the potential for future applications in wearable devices.