Method for stress detection based on wireless resting electroencephalography channel compensation and noise suppression

Abstract

The application discloses a stress detection method based on wireless resting electroencephalogram channel compensation and noise suppression. By constructing a channel compensation mechanism, the spatial structure information of wireless electroencephalogram is recovered, and the feature expression ability under the condition of low channel is improved. Combined with the feature modeling method based on Riemann space, the spatial covariance structure of the electroencephalogram signal is effectively captured, and the stability and discrimination ability of the feature are improved. At the same time, the cross-domain transfer learning strategy is introduced, the feature distribution between the wired electroencephalogram and the wireless electroencephalogram is aligned on the basis of Riemann manifold, and the influence of the wireless electroencephalogram signal noise is reduced. Finally, a unified end-to-end detection framework is constructed, the output of the stress state is realized from the original wireless electroencephalogram signal input, and the accuracy and practical application value of the subclinical depression detection are significantly improved.

Description

Technical Field

[0001] This application belongs to the field of artificial intelligence technology, and in particular relates to a stress detection method based on wireless resting EEG channel compensation and noise suppression. Background Technology

[0002] In recent years, depression has become a significant global public health issue. Subclinical depression (SD), as an early stage of depression development, is characterized by its high reversibility and significant intervention effects. Accurate detection at this stage would help prevent further deterioration of depression. Currently, the assessment of subclinical depression mainly relies on questionnaire scales (such as the Hamilton Depression Rating Scale), but these methods are easily influenced by subjective factors, resulting in biased outcomes and low efficiency, making it difficult to meet the needs for objective and automated detection.

[0003] Electroencephalogram (EEG) signals are widely used in depression detection research due to their non-invasive nature and high temporal resolution. Traditional studies often rely on high-density wired EEG devices for modeling and analysis. While these devices offer high signal quality, they suffer from complexity, high cost, and inconvenience, limiting their widespread application in practical scenarios. In contrast, wireless EEG devices are highly portable and easy to deploy, but they have a limited number of channels, lower signal quality, and are susceptible to motion artifacts, unstable electrode contact, and environmental noise, leading to a significant decline in model performance.

[0004] Although existing studies have used wireless EEG signals to detect depression or psychological stress, the following key problems remain: First, the number of wireless EEG channels is limited, resulting in insufficient spatial information representation and difficulty in fully describing the functional connectivity between brain regions; second, the signal-to-noise ratio of wireless EEG signals is low, and noise interference is severe, leading to poor feature stability; third, there are significant differences in data distribution between wired and wireless EEG, making it difficult to directly transfer traditional models and resulting in insufficient generalization ability, thus limiting the robustness and adaptability of the models in practical applications. Summary of the Invention

[0005] This application provides a stress detection method based on wireless resting EEG channel compensation and noise suppression.

[0006] In a first aspect, embodiments of this application provide a stress detection method based on wireless resting EEG channel compensation and noise suppression, comprising: The acquired raw EEG signals are preprocessed to obtain preprocessed EEG signals; wherein, the raw EEG signals include wired EEG signals and wireless EEG signals, the wired EEG signals include wired task-mode EEG signals and wired resting-mode EEG signals, and the wireless EEG signals include wireless task-mode EEG signals and wireless resting-mode EEG signals; The preprocessed wireless resting-state EEG signal is input into a pre-trained diffusion model to obtain a channel-completed wireless EEG signal, which includes task-state features; wherein, the diffusion model is trained based on the preprocessed wired task-state EEG signal and the wired resting-state EEG signal processed by channel masking. For the preprocessed wired EEG signal and the channel-completed wireless EEG signal, the corresponding covariance matrix is ​​calculated respectively, and the covariance matrix is ​​mapped to the tangent space through Riemannian geometry and then Riemann feature representation is extracted. Using wired EEG signals as the source domain and wireless EEG signals as the target domain, cross-domain feature alignment is performed by using a correlation alignment algorithm and minimizing the Frobenius distance between the Riemann feature representations of the source domain and the Riemann feature representations of the target domain, resulting in domain-aligned features. The domain-aligned features are input into a classifier to obtain the stress detection results.

[0007] Secondly, embodiments of this application provide a stress detection device based on wireless resting EEG channel compensation and noise suppression, comprising: The preprocessing module is used to preprocess the acquired raw EEG signals to obtain preprocessed EEG signals; wherein, the raw EEG signals include wired EEG signals and wireless EEG signals, the wired EEG signals include wired task-mode EEG signals and wired resting-mode EEG signals, and the wireless EEG signals include wireless task-mode EEG signals and wireless resting-mode EEG signals; The completion module is used to input the preprocessed wireless resting-state EEG signal into a pre-trained diffusion model to obtain a channel-completed wireless EEG signal, which includes task-state features; wherein, the diffusion model is trained based on the preprocessed wired task-state EEG signal and the wired resting-state EEG signal processed by channel masking. The spatial feature representation module is used to calculate the corresponding covariance matrix for the preprocessed wired EEG signal and the channel-completed wireless EEG signal, and then extract the Riemann feature representation after mapping the covariance matrix to the tangent space through Riemann geometry. The alignment module is used to take wired EEG signals as the source domain and wireless EEG signals as the target domain, and to perform cross-domain feature alignment by using a correlation alignment algorithm and minimizing the Frobenius distance between the Riemann feature representations of the source domain and the Riemann feature representations of the target domain, so as to obtain the domain-aligned features. The classification module is used to input the domain-aligned features into the classifier to obtain the stress detection results.

[0008] Thirdly, embodiments of this application provide an electronic device comprising: a processor and a memory storing computer program instructions; the processor, when executing the computer program instructions, implements the steps of the stress detection method based on wireless resting EEG channel compensation and noise suppression as described in any embodiment of the first aspect.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the steps of the stress detection method based on wireless resting EEG channel compensation and noise suppression as described in any embodiment of the first aspect.

[0010] Fifthly, embodiments of this application provide a computer program product, which is stored in a storage medium and executed by at least one processor to implement the steps of the stress detection method based on wireless resting EEG channel compensation and noise suppression as provided in the first aspect of embodiments of this application.

[0011] The stress detection method based on wireless resting EEG channel compensation and noise suppression in this application embodiment restores the spatial structure information of wireless EEG by constructing a channel compensation mechanism, thereby improving the feature expression ability under low channel conditions. Combined with a feature modeling method based on Riemann space, it effectively captures the spatial covariance structure of EEG signals, improving the stability and discriminative ability of features. At the same time, it introduces a cross-domain transfer learning strategy to align the feature distribution between wired and wireless EEG based on the Riemann manifold, reducing the impact of noise in wireless EEG signals. Finally, it constructs a unified end-to-end detection framework to realize the output of stress state from the input of raw wireless EEG signals, thereby significantly improving the accuracy and practical application value of subclinical depression detection. Attached Figure Description

[0012] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1This is a schematic flowchart of a stress detection method based on wireless resting EEG channel compensation and noise suppression provided in an embodiment of this application; Figure 2 This is a flowchart illustrating another stress detection method based on wireless resting EEG channel compensation and noise suppression provided in an embodiment of this application. Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0014] The features and exemplary embodiments of various aspects of this application will be described in detail below. To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain this application and not to limit it. For those skilled in the art, this application can be implemented without some of these specific details. The following description of the embodiments is merely to provide a better understanding of this application by illustrating examples.

[0015] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0016] It should be noted that the acquisition, storage, use, and processing of data in this application embodiment all comply with the relevant provisions of national laws and regulations.

[0017] In recent years, depression has become a significant global public health issue. Subclinical depression (SD), as an early stage of depression development, is characterized by its high reversibility and significant intervention effects. Accurate detection at this stage would help prevent further deterioration of depression. Currently, the assessment of subclinical depression mainly relies on questionnaire scales (such as the Hamilton Depression Rating Scale), but these methods are easily influenced by subjective factors, resulting in biased outcomes and low efficiency, making it difficult to meet the needs for objective and automated detection.

[0018] Electroencephalogram (EEG) signals are widely used in depression detection research due to their non-invasive nature and high temporal resolution. Traditional studies often rely on high-density wired EEG devices for modeling and analysis. While these devices offer high signal quality, they suffer from complexity, high cost, and inconvenience, limiting their widespread application in practical scenarios. In contrast, wireless EEG devices are highly portable and easy to deploy, but they have a limited number of channels, lower signal quality, and are susceptible to motion artifacts, unstable electrode contact, and environmental noise, leading to a significant decline in model performance.

[0019] Although existing studies have used wireless EEG signals to detect depression or psychological stress, the following key problems remain: First, the number of wireless EEG channels is limited, resulting in insufficient spatial information representation and difficulty in fully describing the functional connectivity between brain regions; second, the signal-to-noise ratio of wireless EEG signals is low, and noise interference is severe, leading to poor feature stability; third, there are significant differences in data distribution between wired and wireless EEG, making it difficult to directly transfer traditional models and resulting in insufficient generalization ability, thus limiting the robustness and adaptability of the models in practical applications.

[0020] Specifically, the relevant technologies mainly suffer from the following problems: First, the limited number of wireless EEG signal channels leads to spatial information loss. Existing methods often employ channel selection or simple interpolation strategies, which easily result in the loss of key brain region information. This makes it difficult to recover the spatial structural features of high-density EEG under low-channel conditions, affecting the model's ability to model brain network connectivity patterns. Second, the significant difference in domain distribution between wired and wireless EEG signals leads to poor model generalization ability. Compared to wired EEG signals, wireless EEG signals are more susceptible to motion artifacts, unstable electrode contact, and environmental noise during acquisition, resulting in distribution differences between the two. This makes it difficult to fully characterize brain region information related to subclinical depression, affecting detection performance.

[0021] To address the problems in the related technologies, this application provides a stress detection method based on wireless resting EEG channel compensation and noise suppression.

[0022] The stress detection method based on wireless resting EEG channel compensation and noise suppression provided in this application will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0023] Figure 1 A flowchart illustrating the stress detection method based on wireless resting EEG channel compensation and noise suppression, according to an embodiment of this application, is shown. Figure 1 As shown, the stress detection method based on wireless resting EEG channel compensation and noise suppression may specifically include the following steps: S101. The acquired raw EEG signals are preprocessed to obtain preprocessed EEG signals; wherein, the raw EEG signals include wired EEG signals and wireless EEG signals, the wired EEG signals include wired task-mode EEG signals and wired resting-mode EEG signals, and the wireless EEG signals include wireless task-mode EEG signals and wireless resting-mode EEG signals. S102. The preprocessed wireless resting-state EEG signal is input into a pre-trained diffusion model to obtain a channel-completed wireless EEG signal, wherein the channel-completed wireless EEG signal includes task-state features; wherein the diffusion model is trained based on the preprocessed wired task-state EEG signal and the wired resting-state EEG signal processed by channel masking. S103. For the preprocessed wired EEG signal and the channel-completed wireless EEG signal, calculate the corresponding covariance matrix respectively, and then extract the Riemann feature representation after mapping the covariance matrix to the tangent space through Riemann geometry. S104. Using wired EEG signals as the source domain and wireless EEG signals as the target domain, cross-domain feature alignment is performed by using a correlation alignment algorithm and minimizing the Frobenius distance between the Riemann feature representations of the source domain and the Riemann feature representations of the target domain, to obtain the domain-aligned features. S105. Input the domain-aligned features into the classifier to obtain the pressure detection result.

[0024] Therefore, the stress detection method based on wireless resting EEG channel compensation and noise suppression can be applied to the detection of subclinical depression in wireless EEG scenarios with high robustness. It can still extract stable and discriminative EEG features under conditions of low channel number and low signal-to-noise ratio, and achieve effective data transfer across devices.

[0025] The specific implementation methods for each of the above steps are described below.

[0026] In some embodiments, in S101, preprocessing includes at least one of the following: downsampling processing, bandpass filtering processing, rereference processing, physiological artifact removal processing, time segmentation processing, and normalization processing. Specifically, this preprocessing is applied to the acquired wired EEG data and wireless EEG data (including wireless task-mode EEG signals). and wireless resting-state EEG signals ,in Indicates the number of brainwave channels. The data undergoes a standardized preprocessing procedure. First, the data is downsampled. Then, bandpass filtering (0.1–50 Hz) is used to remove low-frequency drift and high-frequency noise. Next, a whole-brain average reference method is used for rereference processing, and independent component analysis is employed to remove physiological artifacts such as electrooculography (EOG) and electromyography (EMG) from the EEG signals. Further, the EEG data is segmented into fixed-length time windows. Finally, the data is normalized to eliminate amplitude differences between different subjects. This systematic preprocessing eliminates noise interference, improves signal quality, and yields structurally uniform and low-noise EEG data, providing reliable input for subsequent modeling.

[0027] As an optional embodiment, before S102, the preprocessed wired resting-state EEG signal is subjected to channel masking to obtain a masked wired resting-state EEG signal; the masked wired resting-state EEG signal is used as the input of the diffusion model, a forward diffusion process is performed on the masked wired resting-state EEG signal and Gaussian noise is gradually added, the preprocessed wired task-state EEG signal is used as a label and reverse denoising is performed to train the diffusion model; it is determined whether the loss value of the diffusion model meets the preset training stopping condition. If it does not meet the condition, the model parameters of the diffusion model are adjusted and the adjusted diffusion model is trained until the preset training stopping condition is met to obtain the trained diffusion model.

[0028] In specific implementation, the preprocessed wired resting-state EEG signal is subjected to channel masking processing based on a channel masking matrix, wherein the channel masking matrix is ​​as follows: , among which, when Time indicates the first One passage was covered, when This indicates that the channel is reserved. This indicates the number of EEG channels. That is, the EEG signal after masking is... , as model input, where This indicates a channel-by-channel multiplication operation.

[0029] In practice, during the training process, the model needs to infer the EEG signals of the masked channels based on the information of the unmasked channels, thereby learning the spatial dependencies between EEG channels. In the diffusion modeling process, a forward diffusion and reverse denoising generation mechanism is adopted. First, in the forward diffusion process, Gaussian noise is gradually added to the input signal to make the data distribution gradually approach the standard Gaussian distribution, as shown in the following formula (1): ; (1) in, Indicates the first Step diffusion state, Indicates the first Step diffusion state, Indicates the first Noise scheduling parameters for each step Indicates the first The noise scheduling parameters for each step are given, where N represents a Gaussian distribution and I represents the identity matrix.

[0030] After multiple diffusion steps, the original EEG signal is gradually covered by noise. Subsequently, in the reverse denoising process, the model directly predicts the EEG signal.

[0031] Optionally, the preset training stopping condition is that the number of times the loss value remains continuously unchanged reaches a preset threshold, wherein the loss value is based on the channel completion loss function. The channel completion loss function is obtained. The mean square error between the model's predicted task-state EEG signals and the actual task-state EEG signals is defined as follows: (2) in, Indicates a time step.

[0032] In this way, by training the diffusion model by simulating channel loss, and by optimizing the model by minimizing the mean square error between the predicted task-state EEG signal and the actual task-state EEG signal, the model's ability to learn spatial information can be enhanced, enabling the trained diffusion model to achieve spatial compensation of EEG signals.

[0033] Furthermore, in some embodiments, in S102, to address the issues of limited channels and missing spatial information in wireless EEG devices, a trained diffusion model is used to recover the complete EEG signal based on partially observable channel information, thereby obtaining a compensated EEG representation. Compared to the original wireless EEG signal, the compensated signal contains richer brain region information in the spatial dimension, providing a more stable input for subsequent Riemannian spatial feature modeling.

[0034] In some embodiments, in S103, firstly, the spatial covariance matrix of each channel-compensated EEG signal is calculated. Then, the covariance matrix is ​​shrunk to obtain the features. To maintain the positive definiteness of the covariance matrix; given the low signal-to-noise ratio and other issues associated with wireless EEG signals, [further details needed]. Diagonal regularization is applied to improve the stability of the covariance estimation. Since the covariance matrix belongs to a symmetric positive definite manifold in non-Euclidean space, and direct feature learning in Euclidean space would destroy its geometric structure, a Riemannian geometric mapping is used to project the covariance matrix into the tangent space, thereby obtaining the feature representation in Euclidean space. Specifically, the logarithmic mapping can be represented by formula (3): (3) in, Represents the reference covariance matrix. This represents a matrix that has undergone regularization.

[0035] Furthermore, for the mapped matrix The vectorization process is performed, and the upper triangular elements are extracted as the final Riemann feature representation. It is understandable that this feature can not only effectively represent the spatial correlation between brain electrical channels, but also preserve the geometric structural information of brain electrical signals.

[0036] In this way, by constructing the covariance matrix of EEG signals and mapping it to a symmetric positive definite manifold space for feature representation, the spatial correlation and functional connectivity between EEG channels can be extracted more effectively.

[0037] In some embodiments, in S104, although channel compensation and Riemann space constraints can enhance the spatial information representation capability of wireless EEG signals, the distribution of wireless EEG signals differs significantly from that of wired EEG signals because wireless EEG devices are more susceptible to motion artifacts, unstable electrode contact, and environmental noise during data acquisition. Therefore, this embodiment introduces an unsupervised adaptive transfer learning mechanism. By aligning the feature distributions between wired and wireless EEG, the model can fully utilize the high-quality information in wired EEG, thereby improving the classification performance on wireless EEG data.

[0038] In practice, wired EEG data is used as the source domain, and wireless EEG data as the target domain. After channel compensation and Riemannian space modeling, Riemannian feature representations of the source and target domains are obtained. Since the two types of data differ in aspects such as acquisition devices and the number of channels, their feature distributions are often inconsistent. Therefore, a domain alignment strategy is used to reduce the distributional differences between the two domains. Given that the covariance matrix of the EEG signal lies in the Riemannian manifold space, a feature alignment method based on correlation alignment (CORAL) is adopted. This method aligns the covariance structures of the source and target domain features, ensuring that the features of the two domains maintain statistical consistency. Subsequently, domain alignment is achieved by minimizing the Frobenius distance between them.

[0039] Therefore, the stress detection method based on wireless resting EEG channel compensation and noise suppression in this application embodiment has at least the following beneficial technical effects: First, compared with existing stress detection or depression detection models that rely on training with a single subject or a single wired EEG signal, it can effectively mitigate the impact of differences in wireless EEG signal distribution across subjects, improve the model's generalization ability and stability among different subjects, and enhance the applicability of the wireless subclinical depression monitoring system in practical application scenarios. Secondly, through the channel compensation mechanism, the spatial information of wireless EEG signals can be effectively restored without increasing the number of hardware channels, enabling low-channel devices to have spatial expression capabilities close to those of high-channel devices, thereby reducing hardware costs and improving data utilization efficiency. Third, by introducing the Riemannian geometric modeling method, it is possible to simultaneously capture the spatial covariance structure of wireless EEG signals and the relationship between the intrinsic geometry of brain regions, avoiding the destruction of EEG structural information by traditional Euclidean space processing, thereby improving the stability and discriminative ability of feature representation. Fourth, through an unsupervised domain adaptive transfer learning mechanism, effective alignment between the distribution of wired and wireless EEG data is achieved, enabling the model to adapt to wireless EEG data based on wired EEG pre-training, thereby improving the robustness and adaptability of the system in wireless EEG device application scenarios.

[0040] In addition, this application also provides another stress detection method based on wireless resting EEG channel compensation and noise suppression. (Reference) Figure 2 This is a flowchart illustrating another stress detection method based on wireless resting EEG channel compensation and noise suppression.

[0041] like Figure 2 As shown, this method is a unified framework that integrates channel compensation, Riemann feature modeling, and transfer learning. Specifically, it consists of a data preprocessing module, a channel completion module, a Riemann constraint module, a domain alignment module, and a classification output layer. The modules are connected in the order of data flow to realize an automated processing flow from raw EEG input to stress feature output.

[0042] The specific implementation of this method may include the following steps: First, the raw EEG signals are preprocessed by the data preprocessing module to eliminate noise interference and improve signal quality, resulting in EEG data with uniform structure and low noise, providing reliable input for subsequent modeling.

[0043] Then, addressing the issues of limited channel count and spatial information loss in wireless EEG devices, a channel completion module is used to mask the input task-state EEG signals during training to simulate channel loss and enhance the model's ability to learn spatial information. In this way, the model infers the EEG signals of masked channels based on the information of unmasked channels during training, thereby learning the spatial dependencies between EEG channels. Thus, the trained diffusion model can recover the complete EEG signal based on partially observable channel information, obtaining a compensated EEG representation. Compared to the original wireless EEG signal, this compensated signal contains richer brain region information in the spatial dimension, providing a more stable input for subsequent Riemannian spatial feature modeling.

[0044] To further enhance the model's ability to model the spatial structure information of EEG signals, a covariance matrix of EEG signals is constructed using a Riemannian space constraint module. This matrix is ​​then mapped to a symmetric positive definite manifold space for feature representation, thereby more effectively extracting the spatial correlation and functional connectivity between EEG channels.

[0045] Although channel completion and Riemann constraint modules can enhance the spatial information representation of wireless EEG signals, the signal distribution of wireless EEG devices differs significantly from that of wired EEG signals due to the increased susceptibility to motion artifacts, unstable electrode contact, and environmental noise during data acquisition. Therefore, relying solely on wireless EEG data for model training can lead to decreased generalization ability. To address this issue, an unsupervised adaptive transfer learning mechanism is introduced. By aligning the feature distributions of wired and wireless EEG data, the model can fully utilize the high-quality information from wired EEG, thereby improving classification performance on wireless EEG data. Specifically, wired EEG data is used as the source domain, and wireless EEG data as the target domain. The Riemann feature representations of the source and target domains are aligned using a feature alignment method based on correlation alignment, ensuring statistical consistency in the feature distributions of the two domains. Domain alignment is achieved by minimizing the Frobenius distance between them. In this way, through the above transfer learning strategy, the model can effectively adapt to wireless EEG data by utilizing the stable feature structure learned from wired EEG data, thereby improving the model's robustness and generalization ability on wireless EEG data.

[0046] In other words, given the significant individual differences in EEG signals across subject scenarios and the distributional offset between wireless and wired EEG data, directly using raw wireless EEG signals for subclinical depression detection often leads to feature instability and insufficient model generalization ability. This application's embodiments introduce channel compensation and transfer learning mechanisms to effectively improve the spatial representation of EEG signals and reduce cross-domain distributional differences without increasing additional hardware costs. Simultaneously, a pre-training approach is used, where the channel compensation model is pre-trained based on wired EEG signals. This module then compensates for the missing spatial information in the wireless EEG signals, making its feature distribution closer to high-quality wired EEG data. Furthermore, unsupervised adaptive transfer learning is used to establish a feature mapping relationship between wired and wireless EEG signals, thereby enhancing the model's stability and consistency under cross-subject conditions and solving the problem of existing methods struggling to maintain high performance in wireless application scenarios.

[0047] In some optional embodiments, the overall optimization objective of the model training corresponding to the overall framework is composed of channel completion loss, domain alignment loss and classification cross-entropy loss, as shown in the following formula (4): (4) in, This represents the cross-entropy loss for classification. = 0.5 is the first weighting coefficient, used to balance the influence between tasks; The domain alignment loss is represented by minimizing the Frobenius distance between the target domain and the source domain to achieve domain alignment. This represents the channel completion loss. Since the channel completion module is a pre-trained module, then here... .

[0048] In this way, by constructing an end-to-end joint optimization framework, channel compensation, Riemann feature constraints, unsupervised domain adaptive transfer learning modules and classification prediction are unified in the same model for collaborative learning. This not only avoids the error accumulation problem in the traditional multi-stage processing flow, but also improves the accuracy, robustness and computational efficiency of subclinical depression detection, thereby enhancing the system's scalability and practical value in cross-subject wireless EEG application scenarios.

[0049] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0050] Based on the same technical concept, and corresponding to any of the above embodiments, this application also provides a pressure detection device based on wireless resting EEG channel compensation and noise suppression.

[0051] Specifically, the stress detection device based on wireless resting EEG channel compensation and noise suppression may include: The preprocessing module is used to preprocess the acquired raw EEG signals to obtain preprocessed EEG signals; wherein, the raw EEG signals include wired EEG signals and wireless EEG signals, the wired EEG signals include wired task-mode EEG signals and wired resting-mode EEG signals, and the wireless EEG signals include wireless task-mode EEG signals and wireless resting-mode EEG signals; The completion module is used to input the preprocessed wireless resting-state EEG signal into a pre-trained diffusion model to obtain a channel-completed wireless EEG signal, which includes task-state features; wherein, the diffusion model is trained based on the preprocessed wired task-state EEG signal and the wired resting-state EEG signal processed by channel masking. The spatial feature representation module is used to calculate the corresponding covariance matrix for the preprocessed wired EEG signal and the channel-completed wireless EEG signal, and then extract the Riemann feature representation after mapping the covariance matrix to the tangent space through Riemann geometry. The alignment module is used to take wired EEG signals as the source domain and wireless EEG signals as the target domain, and to perform cross-domain feature alignment by using a correlation alignment algorithm and minimizing the Frobenius distance between the Riemann feature representations of the source domain and the Riemann feature representations of the target domain, so as to obtain the domain-aligned features. The classification module is used to input the domain-aligned features into the classifier to obtain the stress detection results.

[0052] It should be noted that, for ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0053] The apparatus of the above embodiments is used to implement the corresponding stress detection method based on wireless resting EEG channel compensation and noise suppression in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0054] Based on the same technical concept, corresponding to any of the above embodiments, this application also provides an electronic device.

[0055] Figure 3A schematic diagram of a more specific electronic device hardware structure provided in this embodiment is shown.

[0056] The electronic device 300 may include a processor 301 and a memory 302 storing computer program instructions.

[0057] Specifically, the processor 301 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0058] Memory 302 may include mass storage for data or instructions. For example, and not limitingly, memory 302 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 302 may include removable or non-removable (or fixed) media. Where appropriate, memory 302 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 302 is non-volatile solid-state memory.

[0059] In certain embodiments, the memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Thus, generally, memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to one aspect of this application.

[0060] The processor 301 reads and executes computer program instructions stored in the memory 302 to implement any of the stress detection methods based on wireless resting EEG channel compensation and noise suppression in the above embodiments.

[0061] In some examples, the electronic device 300 may also include a communication interface 303 and a bus 310. For example, Figure 3 As shown, the processor 301, memory 302, and communication interface 303 are connected through bus 310 and complete communication with each other.

[0062] The communication interface 303 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0063] Bus 310 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not as a limitation, bus 310 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 310 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0064] For example, the electronic device 300 can be a mobile phone, tablet computer, laptop computer, handheld computer, in-vehicle electronic device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc.

[0065] Based on the same technical concept, corresponding to any of the methods in the above embodiments, this application also provides a non-transitory computer-readable storage medium. This computer-readable storage medium stores computer program instructions; when executed by a processor, these computer program instructions implement any of the stress detection methods based on wireless resting EEG channel compensation and noise suppression in the above embodiments. Examples of computer-readable storage media include non-transitory computer-readable storage media such as portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, etc.

[0066] Based on the same technical concept, corresponding to any of the above-described embodiments, this application also provides a computer program product, which includes computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processor to perform the stress detection method based on wireless resting EEG channel compensation and noise suppression. Corresponding to the execution entity for each step in each embodiment of the stress detection method based on wireless resting EEG channel compensation and noise suppression, the processor executing the corresponding step can belong to the corresponding execution entity.

[0067] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0068] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0069] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0070] The aspects of this application have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by dedicated hardware performing the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0071] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A stress detection method based on wireless resting EEG channel compensation and noise suppression, characterized in that, include: The acquired raw EEG signals are preprocessed to obtain preprocessed EEG signals; wherein, the raw EEG signals include wired EEG signals and wireless EEG signals, the wired EEG signals include wired task-mode EEG signals and wired resting-mode EEG signals, and the wireless EEG signals include wireless task-mode EEG signals and wireless resting-mode EEG signals; The preprocessed wireless resting-state EEG signal is input into a pre-trained diffusion model to obtain a channel-completed wireless EEG signal, which includes task-state features; wherein, the diffusion model is trained based on the preprocessed wired task-state EEG signal and the wired resting-state EEG signal processed by channel masking. For the preprocessed wired EEG signal and the channel-completed wireless EEG signal, the corresponding covariance matrix is ​​calculated respectively, and the covariance matrix is ​​mapped to the tangent space through Riemannian geometry and then Riemann feature representation is extracted. Using wired EEG signals as the source domain and wireless EEG signals as the target domain, cross-domain feature alignment is performed by using a correlation alignment algorithm and minimizing the Frobenius distance between the Riemann feature representations of the source domain and the Riemann feature representations of the target domain, resulting in domain-aligned features. The domain-aligned features are input into a classifier to obtain the stress detection results.

2. The method according to claim 1, characterized in that, The method further includes, before inputting the preprocessed wireless resting-state EEG signal into a pre-trained diffusion model: The preprocessed wired resting-state EEG signal was subjected to channel masking to obtain the masked wired resting-state EEG signal. Using the masked wired resting-state EEG signal as input to the diffusion model, a forward diffusion process is performed on the masked wired resting-state EEG signal while gradually adding Gaussian noise. The preprocessed wired task-state EEG signal is used as a label and reverse denoising is performed to train the diffusion model. Determine whether the loss value of the diffusion model meets the preset training stopping condition. If not, adjust the model parameters of the diffusion model and train the adjusted diffusion model until the preset training stopping condition is met, thus obtaining the trained diffusion model.

3. The method according to claim 2, characterized in that, The process of performing channel masking on the preprocessed wired resting-state EEG signal to obtain the masked wired resting-state EEG signal includes: Channel masking is performed on the preprocessed wired resting-state EEG signal based on a channel mask matrix, wherein the channel mask matrix is ​​as follows: , among which, when Time indicates the first One passage was covered, when This indicates that the channel is reserved. This indicates the number of brainwave channels.

4. The method according to claim 2, characterized in that, The preset training stopping condition is that the number of times the loss value remains continuously unchanged reaches a preset threshold. The loss value is obtained based on the channel completion loss function, which is the mean square error between the model's predicted task-state EEG signal and the actual task-state EEG signal.

5. The method according to claim 1, characterized in that, The step of extracting Riemann feature representations after mapping the covariance matrix to the tangent space using Riemannian geometry includes: The mapped matrix is ​​vectorized, and its upper triangular elements are extracted as Riemann feature representations.

6. The method according to claim 1, characterized in that, The preprocessing includes at least one of the following: downsampling, bandpass filtering, rereference processing, physiological artifact removal, time segmentation, and normalization.

7. A pressure detection device based on wireless resting EEG channel compensation and noise suppression, characterized in that, include: The preprocessing module is used to preprocess the acquired raw EEG signals to obtain preprocessed EEG signals; wherein, the raw EEG signals include wired EEG signals and wireless EEG signals, the wired EEG signals include wired task-mode EEG signals and wired resting-mode EEG signals, and the wireless EEG signals include wireless task-mode EEG signals and wireless resting-mode EEG signals; The completion module is used to input the preprocessed wireless resting-state EEG signal into a pre-trained diffusion model to obtain a channel-completed wireless EEG signal, which includes task-state features; wherein, the diffusion model is trained based on the preprocessed wired task-state EEG signal and the wired resting-state EEG signal processed by channel masking. The spatial feature representation module is used to calculate the corresponding covariance matrix for the preprocessed wired EEG signal and the channel-completed wireless EEG signal, and then extract the Riemann feature representation after mapping the covariance matrix to the tangent space through Riemann geometry. The alignment module is used to take wired EEG signals as the source domain and wireless EEG signals as the target domain, and to perform cross-domain feature alignment by using a correlation alignment algorithm and minimizing the Frobenius distance between the Riemann feature representations of the source domain and the Riemann feature representations of the target domain, so as to obtain the domain-aligned features. The classification module is used to input the domain-aligned features into the classifier to obtain the stress detection results.

8. An electronic device, characterized in that, The device includes: a processor and a memory storing computer program instructions; when the processor invokes the computer program instructions, it implements the stress detection method based on wireless resting EEG channel compensation and noise suppression as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions, which, when invoked by a processor, implement the stress detection method based on wireless resting EEG channel compensation and noise suppression as described in any one of claims 1-6.

10. A computer program product, characterized in that, When the instructions in the computer program product are executed by the processor of the electronic device, the electronic device performs the stress detection method based on wireless resting EEG channel compensation and noise suppression as described in any one of claims 1-6.

Images