A closed-loop analysis system and method of sound stimuli-brain electrical response

Abstract

The application provides a sound stimulation-brain electrical response closed-loop analysis system and method, relates to the technical field of brain electrical data analysis, and comprises the following steps: obtaining real-time brain electrical data under sound stimulation based on a standardized interface, and performing artifact removal and data standardization operations to obtain standard brain electrical data; performing response latency analysis on the standard brain electrical data to obtain response latency data; performing response significance analysis on the standard brain electrical data to obtain response significance data; extracting an abnormal response event set from the standard brain electrical data, and performing context feature modeling and abnormal root cause analysis on the abnormal response event set based on the response latency data and the response significance data to obtain abnormal root cause data; and generating a brain electrical response analysis graph interface corresponding to the response latency data, the response significance data and the abnormal root cause data based on neural signal analysis logic, and visually displaying the brain electrical response analysis graph interface.

Description

Technical Field

[0001] This invention relates to the field of electroencephalogram (EEG) data analysis technology, specifically a closed-loop analysis system and method for sound stimulation-EEG response. Background Technology

[0002] At the level of EEG response analysis to auditory stimulation, multi-dimensional statistical features were not incorporated to identify latency abnormalities. Response significance analysis often used fixed thresholds, which could not be dynamically adjusted based on historical data, making it difficult to adapt to differences in baseline EEG signals among different subjects and variations in the experimental environment. For root cause analysis of abnormal response events, there was a lack of structured modeling based on multi-dimensional temporal features, making it difficult to quantify the correlation between abnormal features and root causes, resulting in vague root cause localization and poor interpretability. This approach fails to meet the personalized data interpretation needs of researchers, clinicians, and other stakeholders, and cannot efficiently support experimental protocol optimization and clinical intervention decisions.

[0003] Therefore, a closed-loop analysis system and method for sound stimulation-EEG response are provided. Summary of the Invention

[0004] To address the aforementioned technical problems, the present invention aims to provide a closed-loop analysis system and method for sound stimulation-EEG response.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a closed-loop analysis method for sound stimulation-EEG response, the method comprising:

[0006] Real-time EEG data under sound stimulation is acquired using a standardized interface of the EEG acquisition system. Based on the stimulus event identifier, artifact removal and data standardization operations are performed on the real-time EEG data to obtain standard EEG data.

[0007] The standard EEG data was analyzed for response latency based on the trigger time of the sound stimulus and the peak time of the EEG response to obtain response latency data.

[0008] Response significance analysis was performed on the standard EEG data based on the dynamic response intensity threshold to obtain response significance data;

[0009] An abnormal response event set is extracted from the standard EEG data, and context feature modeling and abnormal root cause analysis are performed on the abnormal response event set based on the response latency data and response saliency data to obtain abnormal root cause data.

[0010] Based on neural signal analysis logic, an EEG response analysis graph interface is generated corresponding to the response latency data, response significance data, and abnormal root cause data, and the EEG response analysis graph interface is visualized.

[0011] Furthermore, the process of acquiring real-time EEG data under auditory stimulation based on the standardized interface of the EEG acquisition system includes:

[0012] The EEG acquisition device performs multi-channel signal acquisition in the sound stimulation scenario to obtain raw EEG signals; based on the stimulus presentation logic, the raw EEG signals are correlated with stimulus events and encapsulated to obtain encapsulated EEG data; the encapsulated EEG data is published to a preset neural signal data platform based on the standardized interface of the EEG acquisition system.

[0013] The system receives real-time EEG messages from the neural signal data platform via signal subscription, performs signal parsing on the real-time EEG messages to obtain parsed EEG data, and performs stimulus-response context association on the parsed EEG data to obtain real-time EEG data.

[0014] Furthermore, the process of artifact removal and data standardization of the real-time EEG data based on stimulus event identifiers includes:

[0015] The real-time EEG data is categorized by stimulus type and then tagged with stimulus labels. Based on the stimulus labels, the real-time EEG data is subjected to signal integrity verification. Based on the stimulus event identifier and the signal acquisition time difference threshold, the real-time EEG data after integrity verification is deduplicated to obtain deduplicated EEG data.

[0016] The deduplicated EEG data is filled with missing signals using temporal linear interpolation to obtain completed EEG data. Based on a sliding time window, the completed EEG data is then subjected to EOL and EMG artifact detection to obtain artifact window data. The corresponding artifact components are separated from the valid EEG components using an independent component analysis algorithm. The valid EEG components are retained and the EEG signals are reconstructed to obtain deduplicated EEG data. The deduplicated EEG data is then normalized based on a preset signal amplitude range, and the normalized deduplicated EEG data is further standardized using a preset EEG signal field template to obtain standard EEG data.

[0017] Furthermore, the process of performing response latency analysis on the standard EEG data based on the trigger time of the sound stimulus and the peak time of the EEG response includes:

[0018] Using the stimulus event identifier as the primary key, the corresponding EEG response event group is extracted from the standard EEG data, and then the stimulus triggering event and response peak event are extracted to obtain the trigger-peak event set; the trigger-peak event set is sorted according to the logical order of stimulus presentation to obtain the trigger-peak event sequence set, and the time difference of the corresponding event pairs is calculated to obtain the latency sequence set;

[0019] The latency sequence set is grouped by stimulus type according to the stimulus label to obtain latency group sets; multi-dimensional statistical analysis is performed on each latency group in the latency group set to obtain latency statistical data set; latency anomaly analysis is performed on the latency group set based on the latency statistical data set to obtain an abnormal latency sequence list, and the abnormal latency sequence list and latency statistical data set are used as response latency data.

[0020] Furthermore, the process of performing response significance analysis on the standard EEG data based on a dynamic response intensity threshold includes:

[0021] EEG amplitude sequences are extracted from each EEG response event group to obtain an amplitude sequence set; the amplitude sequence set is sorted according to the stimulus presentation order to obtain an amplitude event sequence set, and the peak intensity sequence set corresponding to the amplitude event sequence set is calculated;

[0022] The peak intensity sequence set is grouped by stimulus type according to the stimulus label to obtain intensity group sets; the historical intensity group sets corresponding to the intensity group sets are obtained, and a corresponding dynamic response intensity threshold is generated for each intensity group based on the historical intensity group sets.

[0023] The intensity groups are filtered for significance based on the dynamic response intensity threshold to obtain significant response groups, and a list of significant response events is generated based on the significant response groups. The significant response proportion of each intensity group in the intensity groups is analyzed based on the significant response groups to obtain a response proportion dataset, and the list of significant response events and the response proportion dataset are used as response significance data.

[0024] Furthermore, the process of extracting an abnormal response event set from the standard EEG data, and performing contextual feature modeling and abnormal root cause analysis on the abnormal response event set based on the response latency data and the response saliency data includes:

[0025] Based on the response latency data and response saliency data, multi-dimensional temporal features are extracted from the standard EEG data to obtain EEG temporal features; based on the EEG temporal features and the abnormal response event set, graph structure modeling and abnormal context modeling are performed on the standard EEG data to obtain EEG association graph structure.

[0026] The abnormal response event set is used as an abnormal label, and the abnormal label is used as a supervision signal to train a neural network model containing a graph attention layer to obtain an EEG attention network model; the EEG association graph structure is forward propagated based on the EEG attention network model to obtain the attention abnormality contribution weight; the attention abnormality contribution weight is subjected to abnormal attribution analysis based on the graph neural network interpreter to obtain abnormal root cause data.

[0027] Furthermore, the process of extracting multi-dimensional temporal features from the standard EEG data based on the response latency data and the response saliency data includes:

[0028] Using the stimulus event identifier as the primary key, the EEG response event groups corresponding to each stimulus event identifier are extracted from the standard EEG data, and the EEG response events in each EEG response event group are sorted according to the stimulus presentation time order to obtain the EEG event sequence set.

[0029] Based on the EEG event sequence set, multi-scale temporal convolution is performed on the response latency data and the response saliency data to obtain latency temporal features and saliency temporal features. An EEG frequency band parameter sequence set is extracted from the EEG event sequence set, and temporal features are extracted from the EEG frequency band parameter sequence set based on a bidirectional long short-term memory network to obtain frequency band temporal features. The latency temporal features, saliency temporal features, and frequency band temporal features are then concatenated to form the EEG temporal features.

[0030] Furthermore, the process of performing graph structure modeling and abnormal context modeling on the standard EEG data based on the EEG temporal features and the abnormal response event set includes:

[0031] The EEG temporal features are used as graph node features to generate an EEG node feature set, and a node edge set corresponding to the EEG node feature set is generated based on the sequence order of the EEG event sequence set. The first few EEG events corresponding to the abnormal response event set in the EEG event sequence set are aggregated into an abnormal pre-event sequence set. Abnormal context modeling is performed on the abnormal response event set based on the response saliency data and response latency data corresponding to the abnormal pre-event sequence set to obtain an abnormal context event sequence set. Abnormal features are embedded into the node edge set based on the abnormal context event sequence set to obtain an edge feature set. An EEG association graph structure is generated based on the EEG node feature set, the node edge set, and the edge feature set.

[0032] Furthermore, the process of generating the EEG response analysis graph interface corresponding to the response latency data, response saliency data, and abnormal root cause data based on neural signal analysis logic includes:

[0033] Using the stimulus event identifier as the primary key, the response latency data, response saliency data, and abnormal root cause data are correlated and integrated to obtain integrated analysis data. Based on the standard EEG data, supplementary event information is added to the integrated analysis data to obtain enhanced analysis data.

[0034] Based on the neural signal analysis logic and the data type of the augmented analysis data, the corresponding chart type is mapped to the augmented analysis data to obtain a chart configuration rule set; based on the preset user role, a Kanban layout template is configured for the chart configuration rule set to obtain a Kanban structure template;

[0035] The enhanced analysis data is instantiated according to the chart configuration rule set to obtain an analysis chart set, and dynamic interaction logic is added to the analysis chart set to obtain a dynamic analysis chart set. The dynamic analysis chart set is then assembled according to the Kanban structure template to obtain an EEG response analysis chart interface, and the EEG response analysis chart interface is then visualized.

[0036] A second aspect of the present invention also provides a closed-loop analysis system for sound stimulation-EEG response, the system comprising a data acquisition module, a data preprocessing module, a data processing module, a data analysis module, and an intelligent visualization module;

[0037] The data acquisition module acquires real-time EEG data under sound stimulation based on the standardized interface of the EEG acquisition system.

[0038] The data preprocessing module performs artifact removal and data standardization operations on the real-time EEG data based on the stimulus event identifiers to obtain standard EEG data.

[0039] The data processing module performs response latency analysis on the standard EEG data based on the trigger time of the sound stimulus and the peak time of the EEG response to obtain response latency data; and performs response significance analysis on the standard EEG data based on the dynamic response intensity threshold to obtain response significance data.

[0040] The data analysis module is used to extract abnormal response event sets from the standard EEG data, and to perform context feature modeling and abnormal root cause analysis on the abnormal response event sets based on the response latency data and response saliency data to obtain abnormal root cause data.

[0041] The intelligent visualization module generates an EEG response analysis graph interface corresponding to the response latency data, response saliency data, and abnormal root cause data based on neural signal analysis logic, and then visualizes the EEG response analysis graph interface.

[0042] Compared with existing technologies, the beneficial effects of this invention are as follows: Real-time EEG data is acquired through a standardized interface of the EEG acquisition system. Artifact removal and data standardization are performed using stimulus event identifiers, effectively solving the problems of inconsistent data formats and low data quality across different devices. This ensures the integrity, consistency, and purity of standard EEG data, providing a high-quality data foundation for subsequent analysis. In the response analysis stage, latency analysis based on the sound stimulus trigger time and the peak time of the EEG response, combined with multi-dimensional statistical features to identify abnormal latencies, improves the comprehensiveness and accuracy of latency analysis. Significance analysis based on dynamically adjusting the response intensity threshold using historical data adapts to differences in different subjects and experimental environments, improving the flexibility and reliability of significant response identification.

[0043] In terms of abnormal root cause analysis, by integrating multi-dimensional temporal characteristics of latency, saliency, and EEG frequency parameters, an EEG association graph structure was constructed. Attribution was then performed using graph attention networks and interpretability models, avoiding the subjectivity of human experience-based judgment and achieving precise localization and quantitative explanation of abnormal root causes. This provides a clear basis for adjusting stimulus protocols and intervening in subject states. In the visualization stage, based on neural signal analysis logic and user-role-customized chart types and dashboard layouts, combined with dynamic interactive functions, complex EEG data was transformed into a visual interface that conforms to professional cognitive habits. This lowered the data interpretation threshold, met the personalized analysis needs of different roles, improved data exploration efficiency and decision support capabilities, and formed a complete closed loop from data collection, processing, analysis to visualization, significantly enhancing the systematicness and practicality of sound stimulus-EEG response analysis. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0045] Figure 1 This is a schematic diagram illustrating the steps of a closed-loop analysis method for sound stimulus-EEG response.

[0046] Figure 2 This is a schematic diagram of a closed-loop analysis system for sound stimulation-EEG response. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0048] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0049] like Figure 1 As shown, a closed-loop analysis method for sound stimulus-EEG response is described, the method comprising the following steps:

[0050] Step S1: Acquire real-time EEG data under sound stimulation based on the standardized interface of the EEG acquisition system, and perform artifact removal and data standardization operations on the real-time EEG data based on the stimulation event identifier to obtain standard EEG data.

[0051] Step S2: Analyze the response latency of the standard EEG data based on the trigger time of the sound stimulus and the peak time of the EEG response to obtain response latency data;

[0052] Step S3: Perform response significance analysis on the standard EEG data based on the dynamic response intensity threshold to obtain response significance data;

[0053] Step S4: Extract the abnormal response event set from the standard EEG data, and perform context feature modeling and abnormal root cause analysis on the abnormal response event set based on the response latency data and response saliency data to obtain abnormal root cause data.

[0054] Step S5: Generate an EEG response analysis graph interface corresponding to the response latency data, response significance data, and abnormal root cause data based on the neural signal analysis logic, and visualize the EEG response analysis graph interface.

[0055] It should be further explained that, in the specific implementation process, the process of acquiring real-time EEG data under sound stimulation based on the standardized interface of the EEG acquisition system includes:

[0056] Optionally, in this embodiment of the application, the EEG acquisition device under sound stimulation scenario performs multi-channel signal acquisition to obtain raw EEG signals;

[0057] Based on the stimulus presentation logic, the original EEG signals are correlated with stimulus events and encapsulated to obtain encapsulated EEG data.

[0058] The encapsulated EEG data is published to a pre-defined neural signal data platform based on the standardized interface of the EEG acquisition system.

[0059] The system receives real-time EEG messages from the neural signal data platform via signal subscription and performs signal analysis on the real-time EEG messages to obtain analyzed EEG data.

[0060] The analyzed EEG data is subjected to stimulus-response context association to obtain real-time EEG data.

[0061] It should be noted that the EEG acquisition system includes an EEG instrument for acquiring EEG signals, such as an electroencephalogram (EEG) machine, as well as a portable EEG acquisition device, a stimulus presentation device, and a data transmission control system. The EEG acquisition system is an EEG signal acquisition and processing system for neuroscience research or clinical testing. It achieves precise matching between sound stimulation and EEG response through multi-channel synchronous acquisition, recording the complete response process from sound stimulus presentation to EEG signal generation, conduction, and attenuation. The number of channels of the EEG acquisition device can be configured according to needs, such as 32 channels, 64 channels, or 128 channels. The acquisition frequency is not lower than a preset frequency threshold to ensure the capture of high-frequency EEG components.

[0062] To further explain, the multi-channel signal acquisition refers to the synchronous acquisition of raw EEG signals from various regions of the scalp using the electrode array of the EEG acquisition device. These raw EEG signals include basic data such as EEG amplitude, acquisition timestamp, electrode number, and channel gain. Simultaneously, stimulus presentation-related data is collected. This stimulus presentation-related data includes, but is not limited to, stimulus type (e.g., peaceful sound stimuli such as rustling leaves, the gentle flow of a stream, or a slow, even piano melody), negative sound stimuli (e.g., car horns, drill sounds, high-frequency mechanical noises, or metallic friction sounds), or positive sound stimuli (melodic pop songs, baby laughter, or cheering shouts), and stimulus frequency (e.g., 20Hz-). The stimulation intensity (e.g., 40dB-80dB) and duration (e.g., 10ms-1000ms) are specified. The stimulus presentation logic refers to the correspondence between the presentation order, interval, and repetition count of the sound stimuli and the EEG acquisition channels. For example, pure tone stimuli are presented cyclically at frequency gradients of 1kHz, 2kHz, and 4kHz, with a 2-second interval between each stimulus, and each frequency is repeated 10 times. The stimulus event association refers to associating the presentation event of each sound stimulus with the EEG acquisition data within the corresponding time window. These stimulus events can be stimulus start events, stimulus end events, stimulus interval events, and stimulus repetition events, etc. The standardized interface can... The interface used is either the Brain Imaging Data Structure (BIDS) or the Digital Imaging and Communications in Medicine (DICOM) interface. The neural signal data platform is a dedicated platform for storing, managing, and distributing EEG and stimulation data, supporting multi-source data fusion and real-time transmission. The signal subscription method can be a real-time communication protocol such as Data Distribution Service (DDS) or Message Queuing Telemetry Transport (MQTT). Signal parsing refers to extracting the data format identifier from the protocol header of the real-time EEG message, and extracting the parsed EEG data from the message body based on the data format identifier. The parsed EEG data includes, but is not limited to, EEG amplitude sequences, timestamp sequences, and channel configuration information. Stimulus response context association refers to associating the parsed EEG data with the subject information (e.g., age, gender, and health status), experimental scenario information (e.g., laboratory environment and noise level), and experimental protocol data (e.g., stimulus sequence and task requirements) of the neural signal data platform, assigning it complete stimulus response context information.

[0063] It should be further explained that, in the specific implementation process, the process of performing artifact removal and data standardization on the real-time EEG data based on stimulus event identifiers to obtain standard EEG data includes:

[0064] Optionally, in this embodiment of the application, the real-time EEG data is classified by stimulation type, and stimulation tags are added to the real-time EEG data based on the result of stimulation type classification;

[0065] The real-time EEG data is subjected to signal integrity verification based on stimulus tags, and the real-time EEG data after integrity verification is deduplicated based on stimulus event identifiers and signal acquisition time difference thresholds to obtain deduplicated EEG data.

[0066] The deduplicated EEG data was filled with missing signals using a time-series linear interpolation method to obtain the completed EEG data.

[0067] Based on the sliding time window, the complete EEG data is subjected to EEG artifact detection and EMG artifact detection to obtain artifact window data. In the sliding time window, the artifact component and the valid EEG component are separated in the artifact window data using the independent component analysis algorithm.

[0068] The effective EEG components are retained and the EEG signals are reconstructed to obtain artifact-free EEG data.

[0069] The artifact-free EEG data is normalized based on a preset signal amplitude range, and then the normalized artifact-free EEG data is standardized based on a preset EEG signal field template to obtain standard EEG data.

[0070] It should be noted that the stimulation type classification refers to determining the stimulation type corresponding to each data point based on the stimulation presentation-related data in the real-time EEG data. The stimulation types include, but are not limited to, pure tone stimulation, complex tone stimulation, speech stimulation, and music stimulation. The stimulation label is a unique identifier for each type of stimulation. The signal integrity verification refers to checking whether the real-time EEG data contains all necessary attributes for each stimulation type, such as the corresponding stimulation presentation-related data, EEG amplitude, timestamp, and channel data. If so, the real-time EEG data is complete; otherwise, the missing attributes are considered incomplete data. The stimulation event identifier is a unique identifier corresponding to each sound stimulation event, for example, the stimulation sequence number S20240520-001-003. The data deduplication refers to deduplication based on the repetition of stimulus event identifiers and the time difference between two consecutive identical data acquisitions under the same stimulus event identifier. Specifically, when the identifiers of two stimulus events are repeated, the data is determined to be duplicated and deduplicated. Under the same stimulus event identifier, if the time difference between two identical data acquisitions of the same channel and type is less than a preset signal acquisition time difference threshold, such as 1ms, the data is determined to be duplicated and deduplicated.

[0071] To further explain, the method of using temporal linear interpolation to complete the missing signals in the deduplicated EEG data refers to filtering out missing channel data or time point data from the deduplicated EEG data, generating complete EEG data based on the temporal position of the missing data and using the linear relationship between adjacent valid data points to ensure the continuity of the temporal data; setting the window size and step size for the sliding time window, for example, setting the window size to 200ms-500ms and the step size to 50ms-100ms, and separating artifact components using the Independent Component Analysis (ICA) algorithm, specifically decomposing the EEG signal into multiple independent components, identifying oculomotor artifacts (e.g., low-frequency high-amplitude components generated by blinking) and electromyographic artifacts (e.g., high-frequency components generated by facial muscle contraction) through the spectral characteristics and spatial distribution characteristics of the components, thereby obtaining the corresponding artifact components, removing the artifact components from the original signal, and reconstructing the valid EEG signal.

[0072] It should be noted that acquiring real-time EEG data through the standardized interface of the EEG acquisition system enables precise synchronization between stimulus presentation and signal acquisition, ensuring the real-time nature and relevance of the data. Stimulus labeling, integrity verification, and time difference threshold deduplication avoid duplicate acquisition and interference from invalid data. Time-series linear interpolation completes missing data, and the ICA algorithm removes artifacts, significantly improving the purity and continuity of the EEG data. Finally, normalization and structural standardization provide a high-quality, uniformly formatted data foundation for subsequent latency analysis, significance analysis, and abnormal root cause analysis.

[0073] It should be further explained that, in the specific implementation process, the process of performing response latency analysis on the standard EEG data based on the trigger time of the sound stimulus and the peak time of the EEG response to obtain the response latency data includes:

[0074] Optionally, in this embodiment of the application, the stimulus event identifier is used as the primary key to extract the EEG response event group corresponding to each stimulus event identifier from the standard EEG data, and the stimulus trigger event and response peak event are extracted from each EEG response event group to obtain the trigger-peak event group set;

[0075] The trigger-peak event set is sorted according to the logical order of stimulus presentation to obtain the trigger-peak event sequence set, and the time difference between each stimulus trigger event and response peak event pair in the trigger-peak event sequence set is calculated to obtain the latency sequence set;

[0076] The latency sequence set is grouped by stimulus type according to the stimulus label to obtain latency group sets;

[0077] Multidimensional statistical analysis is performed on each latent period group in the latent period group set to obtain a latent period statistical data set, wherein the latent period statistical data set includes a set of latent period mean, a set of latent period median, a set of latent period standard deviation, and a set of latent period fluctuation coefficient;

[0078] Based on the incubation period statistical data set, an incubation period anomaly analysis is performed on the incubation period set to obtain an abnormal incubation sequence list, and the abnormal incubation sequence list and the incubation period statistical data set are used as response incubation period data.

[0079] It should be noted that the EEG response event group refers to all EEG response data corresponding to a complete stimulus event, including pre-stimulation baseline data, during-stimulation response data, and post-stimulation recovery data. The stimulus triggering event refers to the time point at which the auditory stimulus begins to be presented, i.e., the corresponding timestamp. The response peak event refers to the time point at which the EEG signal amplitude reaches its local maximum after stimulus presentation, such as the peak time of the P300 component and the trough time of the N100 component. The calculation of the latency sequence set refers to calculating the difference between the timestamp of each stimulus triggering event and the timestamp of the corresponding response peak event to obtain the response latency of a single stimulus event; for example, 300ms indicates that the EEG peak response occurs 300ms after stimulus presentation. The latency anomaly analysis refers to identifying abnormal latencies based on the statistical characteristics of each latency group; for example, selecting latency groups one by one as target groups and calculating the mean latency of that group. with standard deviation , will be greater than or less The latency period was determined to be an abnormal latency period, and the stimulus events, subject information, and stimulus presentation data corresponding to the abnormal latency period were compiled into an abnormal latency sequence list.

[0080] It should be noted that by aggregating EEG response data through stimulus event identification, the complete response process corresponding to each stimulus can be accurately located, ensuring the relevance of latency calculation; by grouping and statistically analyzing by stimulus type, the latency distribution characteristics under different stimulus patterns can be obtained, reflecting the differences in response speed of subjects to different sound stimuli; through multi-dimensional statistics and anomaly identification, key features of the latency dimension are provided for subsequent abnormal root cause analysis.

[0081] It should be further explained that, in the specific implementation process, the process of performing response significance analysis on the standard EEG data based on the dynamic response intensity threshold to obtain response significance data includes:

[0082] Optionally, in the embodiments of this application, EEG amplitude sequences are extracted from each EEG response event group to obtain an amplitude sequence set;

[0083] The amplitude sequence set is sorted according to the stimulus presentation order to obtain the amplitude event sequence set, and the peak intensity sequence set corresponding to the amplitude event sequence set is calculated.

[0084] The peak intensity sequence set is grouped by stimulus type according to the stimulus label to obtain intensity groups;

[0085] Obtain the historical intensity set corresponding to the intensity set, and generate a corresponding dynamic response intensity threshold for each intensity set based on the historical intensity set;

[0086] The intensity set is filtered for significance based on the dynamic response intensity threshold to obtain a significant response set, and a list of significant response events is generated based on the significant response set.

[0087] Based on the significant response set, a significant response proportion analysis is performed on each intensity group in the intensity set to obtain a response proportion dataset. The list of significant response events and the response proportion dataset are used as response significance data.

[0088] It should be noted that the EEG amplitude sequence refers to the sequence of EEG signal amplitude changes over time corresponding to a single stimulus event. The peak intensity sequence set refers to the sequence composed of peak amplitudes extracted from each amplitude sequence, such as the maximum positive amplitude or the maximum negative amplitude. The historical intensity set refers to the set of peak intensity data formed in historical experiments with the same stimulus type, the same subjects, or the same subject group. The generation of a dynamic response intensity threshold based on the historical intensity set refers to calculating the mean of each historical intensity set. with standard deviation ,Will As a dynamic response intensity threshold for the corresponding stimulus type, the dynamic response intensity threshold is dynamically adjusted as historical data is updated. The saliency screening refers to determining peak intensities in intensity groups that exceed the corresponding dynamic response intensity threshold as saliency responses, and compiling the stimulus events, peak amplitudes, and response times corresponding to saliency responses into a saliency response event list. The saliency response proportion analysis refers to calculating the proportion of saliency responses in each intensity group to the total number of responses in that group, as well as the proportion of saliency responses in different amplitude ranges. For example… , .

[0089] Furthermore, dynamically adjusting the response intensity threshold using historical data can adapt to differences in the baseline of EEG signals among different subjects and changes in the experimental environment, thereby improving the flexibility and accuracy of significance analysis. By distinguishing the intensity characteristics of different stimulus types, it is possible to accurately identify the sensitive response patterns of subjects to specific sound stimuli. The percentage of significant responses can reflect the stability of stimulus responses, providing key evidence for the significance dimension of abnormal root cause analysis.

[0090] It should be further explained that, in the specific implementation process, the process of extracting the abnormal response event set from the standard EEG data, and performing contextual feature modeling and abnormal root cause analysis on the abnormal response event set based on the response latency data and the response saliency data to obtain abnormal root cause data includes:

[0091] Optionally, in the embodiments of this application, the abnormal response event set refers to a set of abnormal signal events that occur during the EEG response process. The abnormal response events include, but are not limited to, amplitude abnormal events (e.g., amplitude exceeding the normal range by more than 10 times), latency abnormal events (e.g., latency exceeding the average latency of similar stimuli by more than 3 times), waveform distortion events (e.g., response waveform without obvious peaks), and no response events (e.g., no effective response detected after stimulation). The abnormal response events include an abnormality type label, such as amplitude overload, latency delay, waveform distortion, or no response.

[0092] Optionally, in this embodiment of the application, multi-dimensional temporal features are extracted from the standard EEG data based on the response latency data and the response saliency data to obtain EEG temporal features;

[0093] Based on the EEG temporal features and the abnormal response event set, graph structure modeling and abnormal context modeling are performed on the standard EEG data to obtain the EEG association graph structure.

[0094] The abnormal response event set is used as an abnormal label, and the abnormal label is used as a supervision signal to train a neural network model containing a graph attention layer, thereby obtaining an EEG attention network model.

[0095] Based on the EEG attention network model, the EEG correlation graph structure is forward-propagated to obtain the contribution weight of attention abnormality.

[0096] Anomaly attribution analysis is performed on the attention anomaly contribution weights based on a graph neural network interpreter to obtain anomaly root cause data.

[0097] It should be noted that, since the response latency data reflects the temporal characteristics of the EEG response and the response saliency data reflects the intensity characteristics of the EEG response, both are directly related to abnormal response events. For example, latency delay may lead to no response, and amplitude abnormality may cause waveform distortion. Therefore, the EEG temporal features extracted by combining the two can comprehensively characterize the contextual information of abnormal responses and improve the accuracy of abnormal attribution.

[0098] To further explain, when training the neural network model containing the graph attention layer, the cross-entropy loss function or the mean squared error loss function is used for model optimization. Simultaneously, a combination of oversampling and undersampling is used to address class imbalance, ensuring the model's sensitivity in identifying anomalous samples. The neural network model uses Graph-LRP, a higher-order interpretation of graph neural networks based on relevant paths. Backpropagation is used to calculate the contribution of each node's features to the anomalous result, quantifying the influence weights of latency anomalies, saliency anomalies, or frequency band parameter anomalies on the final anomalous response. The anomaly attribution analysis refers to determining the core cause of the anomalous response based on the attention anomaly contribution weights and the output results of the neural network model, combined with knowledge from the field of neuroscience. For example, if the attention weight of a non-response event is found to be mainly concentrated on a sudden drop in the amplitude of the beta band of the preceding stimulus, and the time interval between the preceding stimulus and the non-response event is less than 500ms, it can be attributed to the blockage of neural response conduction caused by beta band activity inhibition; if the weight of an amplitude overload event is concentrated on the stimulus intensity exceeding the subject's tolerance threshold, it can be attributed to excessive neural activation caused by unreasonable stimulus parameter settings.

[0099] To further explain, the aforementioned EEG correlation graph structure can comprehensively capture the dynamic correlations and abnormal precursors during the stimulus response process; by using abnormal events as supervisory signals to train the graph attention network, the model can autonomously learn the weight distribution of abnormal features, avoiding the subjectivity of manual feature selection; and by combining an interpretability model for abnormal attribution, it can not only locate the root cause of the abnormality, but also output the attribution basis, providing practical guidance and suggestions for subsequent stimulus program adjustments or subject state intervention.

[0100] It should be further explained that, in the specific implementation process, the process of extracting multi-dimensional temporal features from the standard EEG data based on the response latency data and the response saliency data to obtain the EEG temporal features includes:

[0101] Optionally, in this embodiment of the application, the stimulus event identifier is used as the primary key to extract the EEG response event group corresponding to each stimulus event identifier from the standard EEG data, and the EEG response events in each EEG response event group are sorted according to the stimulus presentation time order to obtain an EEG event sequence set.

[0102] Based on the EEG event sequence set, multi-scale temporal convolution is performed on the response latency data and the response saliency data to obtain latency temporal features and saliency temporal features;

[0103] A set of EEG frequency band parameter sequences is extracted from the set of EEG event sequences. The set of EEG frequency band parameter sequences includes alpha band amplitude sequences, beta band amplitude sequences, theta band amplitude sequences, and delta band amplitude sequences. Temporal features are extracted from the set of EEG frequency band parameter sequences based on a bidirectional long short-term memory network to obtain frequency band temporal features.

[0104] The latency time-series features, saliency time-series features, and frequency band time-series features are concatenated to form EEG time-series features.

[0105] It should be noted that the multi-scale temporal convolution refers to using convolution kernels of different sizes (e.g., convolution kernels corresponding to 3ms, 5ms, and 7ms time windows) to perform convolution operations on latency data and saliency data, capturing feature patterns at different time scales to obtain high-dimensional temporal feature vectors; the EEG frequency band parameter sequence refers to decomposing the EEG amplitude sequence into amplitude sequences of different frequency bands through Fast Fourier Transform (FFT). The activity of the corresponding frequency band is closely related to cognitive function and emotional state, and abnormal changes in these bands may lead to abnormal EEG responses. Therefore, the frequency band temporal features can reflect the correlation between abnormal responses and brain functional states.

[0106] It should be further explained that, in the specific implementation process, the specific process of performing graph structure modeling and abnormal context modeling on the standard EEG data based on the EEG temporal features and the abnormal response event set to obtain the EEG association graph structure includes:

[0107] Optionally, in this embodiment of the application, the EEG temporal features are used as graph node features to generate an EEG node feature set, and the node edge set corresponding to the EEG node feature set is generated based on the sequence order of the EEG event sequence set.

[0108] The first few EEG events corresponding to the abnormal response event set in the EEG event sequence set are aggregated into an abnormal pre-event sequence set. Based on the response saliency data and response latency data corresponding to the abnormal pre-event sequence set, the abnormal response event set is modeled with an abnormal context to obtain an abnormal context event sequence set.

[0109] Based on the abnormal context event sequence set, abnormal features are embedded into the node edge set to obtain the edge feature set, and an EEG association graph structure is generated based on the EEG node feature set, the node edge set, and the edge feature set.

[0110] It should be noted that the sequence order refers to the temporal order of the sound stimuli and the corresponding EEG response order. Each graph node corresponds to the EEG temporal characteristics of a stimulus event, and the node edge set reflects the temporal correlation between adjacent stimulus events. The first few EEG events can be set as the first 3 to 5 events. By analyzing the latency and saliency data before the occurrence of abnormal responses, temporal abnormal precursor features can be discovered. The abnormal context modeling refers to labeling the latency fluctuations and saliency changes of each abnormal response event with its preceding events, and recording the time interval and stimulus parameter differences between the preceding events and the abnormal events to enhance feature correlation. The abnormal feature embedding refers to converting abnormal context information into vector form as the features of node edges, so that the EEG correlation graph structure not only contains node features, but also can characterize the abnormal correlation between events through edge features, providing richer contextual information for the feature learning of the subsequent graph attention network.

[0111] It should be further explained that, in the specific implementation process, the process of generating the EEG response analysis graph interface corresponding to the response latency data, response saliency data, and abnormal root cause data based on neural signal analysis logic, and visually displaying the EEG response analysis graph interface, includes:

[0112] It should be noted that after calculating the response latency data, response significance data, and abnormal root cause data, in order to intuitively present the correlation, abnormal distribution, and root cause relationship between sound stimulation and EEG response, and to facilitate researchers or clinicians to quickly grasp the experimental results and locate key issues, it is necessary to generate a corresponding EEG response analysis interface.

[0113] Optionally, in this embodiment of the application, the stimulus event identifier is used as the primary key to associate and integrate the response latency data, response saliency data, and abnormal root cause data to obtain integrated analysis data, and supplementary event information is added to the integrated analysis data based on the standard EEG data to obtain enhanced analysis data;

[0114] Based on the neural signal analysis logic and the data type of the augmented analysis data, the corresponding chart type is mapped to the augmented analysis data to obtain the chart configuration rule set;

[0115] Configure the Kanban layout template for the chart based on the preset user roles and the rule set of the chart configuration to obtain the Kanban structure template;

[0116] The enhanced analysis data is instantiated according to the chart configuration rule set to obtain an analysis chart set, and dynamic interaction logic is added to the analysis chart set to obtain a dynamic analysis chart set.

[0117] The interface of the dynamic analysis chart set is assembled according to the Kanban structure template to obtain the EEG response analysis chart interface, and the EEG response analysis chart interface is visualized. The specific visualization process will not be described in detail.

[0118] It should be noted that the supplementary event information refers to adding the subject's real-time physiological state, environmental parameters at the time of stimulus presentation, and spatial location information of the EEG acquisition channel. This information can help analyze the correlation between abnormal responses and the external environment or physiological state. The neural signal analysis logic refers to the conventional analysis dimensions of EEG data in neuroscience research or clinical testing, such as analyzing the differences in latency distribution of different stimulus types, assessing the correlation between specific frequency band parameters and response significance, and tracing the root cause transmission pathway of abnormal responses.

[0119] like Figure 2 As shown, a closed-loop analysis system for sound stimulation-EEG response is provided, the system including but not limited to a data acquisition module, a data preprocessing module, a data processing module, a data analysis module and an intelligent visualization module;

[0120] The data acquisition module acquires real-time EEG data under sound stimulation based on the standardized interface of the EEG acquisition system.

[0121] The data preprocessing module performs artifact removal and data standardization operations on the real-time EEG data based on the stimulus event identifiers to obtain standard EEG data.

[0122] The data processing module performs response latency analysis on the standard EEG data based on the trigger time of the sound stimulus and the peak time of the EEG response to obtain response latency data; and performs response significance analysis on the standard EEG data based on the dynamic response intensity threshold to obtain response significance data.

[0123] The data analysis module is used to extract abnormal response event sets from the standard EEG data, and to perform context feature modeling and abnormal root cause analysis on the abnormal response event sets based on the response latency data and response saliency data to obtain abnormal root cause data.

[0124] The intelligent visualization module generates an EEG response analysis graph interface corresponding to the response latency data, response saliency data, and abnormal root cause data based on neural signal analysis logic, and then visualizes the EEG response analysis graph interface.

[0125] Optionally, in this embodiment, those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0126] The sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0127] If the integrated units in the above embodiments are implemented as software functional units and sold or used as independent products, they can be stored in the aforementioned computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause one or more electronic devices to execute all or part of the steps of the methods described in the various embodiments of this application.

[0128] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0129] In the several embodiments provided in this application, it should be understood that the disclosed application can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of units or modules may be electrical or other forms.

[0130] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0131] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0132] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A closed-loop analysis method for sound stimulus-EEG response, characterized in that, The method includes: Real-time EEG data under sound stimulation is acquired using a standardized interface of the EEG acquisition system. Based on the stimulus event identifier, artifact removal and data standardization operations are performed on the real-time EEG data to obtain standard EEG data. The standard EEG data was analyzed for response latency based on the trigger time of the sound stimulus and the peak time of the EEG response to obtain response latency data. Response significance analysis was performed on the standard EEG data based on the dynamic response intensity threshold to obtain response significance data; An abnormal response event set is extracted from the standard EEG data, and contextual feature modeling and abnormal root cause analysis are performed on the abnormal response event set based on the response latency data and response saliency data. The process of performing context feature modeling and root cause analysis on the abnormal response event set includes: Based on the response latency data and response saliency data, multi-dimensional temporal features are extracted from the standard EEG data to obtain EEG temporal features; based on the EEG temporal features and the abnormal response event set, graph structure modeling and abnormal context modeling are performed on the standard EEG data to obtain EEG association graph structure. The abnormal response event set is used as an anomaly label, and the anomaly label is used as a supervision signal to train a neural network model containing a graph attention layer to obtain an EEG attention network model; the EEG association graph structure is forward-propagated based on the EEG attention network model to obtain the attention anomaly contribution weight; the attention anomaly contribution weight is subjected to anomaly attribution analysis based on the graph neural network interpreter to obtain the anomaly root cause data. The process of extracting multi-dimensional temporal features from the standard EEG data includes: Using the stimulus event identifier as the primary key, the EEG response event groups corresponding to each stimulus event identifier are extracted from the standard EEG data, and the EEG response events in each EEG response event group are sorted according to the stimulus presentation time order to obtain the EEG event sequence set. Based on the EEG event sequence set, multi-scale temporal convolution is performed on the response latency data and the response saliency data to obtain latency temporal features and saliency temporal features; an EEG frequency band parameter sequence set is extracted from the EEG event sequence set, and temporal features are extracted from the EEG frequency band parameter sequence set based on a bidirectional long short-term memory network to obtain frequency band temporal features; the latency temporal features, saliency temporal features, and frequency band temporal features are concatenated to form EEG temporal features; Based on neural signal analysis logic, an EEG response analysis graph interface is generated corresponding to the response latency data, response significance data, and abnormal root cause data, and the EEG response analysis graph interface is visualized.

2. The closed-loop analysis method for sound stimulus-EEG response according to claim 1, characterized in that, The process of acquiring real-time EEG data under auditory stimulation using a standardized interface of an EEG acquisition system includes: The EEG acquisition device performs multi-channel signal acquisition in the sound stimulation scenario to obtain raw EEG signals; based on the stimulus presentation logic, the raw EEG signals are correlated with stimulus events and encapsulated to obtain encapsulated EEG data; the encapsulated EEG data is published to a preset neural signal data platform based on the standardized interface of the EEG acquisition system. The system receives real-time EEG messages from the neural signal data platform via signal subscription, performs signal parsing on the real-time EEG messages to obtain parsed EEG data, and performs stimulus-response context association on the parsed EEG data to obtain real-time EEG data.

3. The closed-loop analysis method for sound stimulus-EEG response according to claim 2, characterized in that, The process of artifact removal and data standardization of the real-time EEG data based on stimulus event identifiers includes: The real-time EEG data is categorized by stimulus type and then tagged with stimulus labels. Based on the stimulus labels, the real-time EEG data is subjected to signal integrity verification. Based on the stimulus event identifier and the signal acquisition time difference threshold, the real-time EEG data after integrity verification is deduplicated to obtain deduplicated EEG data. The deduplicated EEG data is filled with missing signals using temporal linear interpolation to obtain completed EEG data. Based on a sliding time window, the completed EEG data is then subjected to EOL and EMG artifact detection to obtain artifact window data. The corresponding artifact components are separated from the valid EEG components using an independent component analysis algorithm. The valid EEG components are retained and the EEG signals are reconstructed to obtain deduplicated EEG data. The deduplicated EEG data is then normalized based on a preset signal amplitude range, and the normalized deduplicated EEG data is further standardized using a preset EEG signal field template to obtain standard EEG data.

4. The closed-loop analysis method for sound stimulus-EEG response according to claim 3, characterized in that, The process of analyzing the response latency of the standard EEG data based on the trigger time of the sound stimulus and the peak time of the EEG response includes: Using the stimulus event identifier as the primary key, the corresponding EEG response event group is extracted from the standard EEG data, and then the stimulus triggering event and response peak event are extracted to obtain the trigger-peak event set; the trigger-peak event set is sorted according to the logical order of stimulus presentation to obtain the trigger-peak event sequence set, and the time difference of the corresponding event pairs is calculated to obtain the latency sequence set; The latency sequence set is grouped by stimulus type according to the stimulus label to obtain latency group sets; multi-dimensional statistical analysis is performed on each latency group in the latency group set to obtain latency statistical data set; latency anomaly analysis is performed on the latency group set based on the latency statistical data set to obtain an abnormal latency sequence list, and the abnormal latency sequence list and latency statistical data set are used as response latency data.

5. The closed-loop analysis method for sound stimulus-EEG response according to claim 4, characterized in that, The process of performing response significance analysis on the standard EEG data based on the dynamic response intensity threshold includes: EEG amplitude sequences are extracted from each EEG response event group to obtain an amplitude sequence set; the amplitude sequence set is sorted according to the stimulus presentation order to obtain an amplitude event sequence set, and the peak intensity sequence set corresponding to the amplitude event sequence set is calculated; The peak intensity sequence set is grouped by stimulus type according to the stimulus label to obtain intensity group sets; the historical intensity group sets corresponding to the intensity group sets are obtained, and a corresponding dynamic response intensity threshold is generated for each intensity group based on the historical intensity group sets. The intensity groups are filtered for significance based on the dynamic response intensity threshold to obtain significant response groups, and a list of significant response events is generated based on the significant response groups. The significant response proportion of each intensity group in the intensity groups is analyzed based on the significant response groups to obtain a response proportion dataset, and the list of significant response events and the response proportion dataset are used as response significance data.

6. The closed-loop analysis method for sound stimulus-EEG response according to claim 5, characterized in that, The process of performing graph structure modeling and abnormal context modeling on the standard EEG data based on the EEG temporal features and the abnormal response event set includes: The EEG temporal features are used as graph node features to generate an EEG node feature set, and a node edge set corresponding to the EEG node feature set is generated based on the sequence order of the EEG event sequence set. The first few EEG events corresponding to the abnormal response event set in the EEG event sequence set are aggregated into an abnormal pre-event sequence set. Abnormal context modeling is performed on the abnormal response event set based on the response saliency data and response latency data corresponding to the abnormal pre-event sequence set to obtain an abnormal context event sequence set. Abnormal features are embedded into the node edge set based on the abnormal context event sequence set to obtain an edge feature set. An EEG association graph structure is generated based on the EEG node feature set, the node edge set, and the edge feature set.

7. The closed-loop analysis method for sound stimulus-EEG response according to claim 6, characterized in that, The process of generating the EEG response analysis graph interface corresponding to the response latency data, response significance data, and abnormal root cause data based on neural signal analysis logic includes: Using the stimulus event identifier as the primary key, the response latency data, response saliency data, and abnormal root cause data are correlated and integrated to obtain integrated analysis data. Based on the standard EEG data, supplementary event information is added to the integrated analysis data to obtain enhanced analysis data. Based on the neural signal analysis logic and the data type of the augmented analysis data, the corresponding chart type is mapped to the augmented analysis data to obtain a chart configuration rule set; based on the preset user role, a Kanban layout template is configured for the chart configuration rule set to obtain a Kanban structure template; The enhanced analysis data is instantiated according to the chart configuration rule set to obtain an analysis chart set, and dynamic interaction logic is added to the analysis chart set to obtain a dynamic analysis chart set. The dynamic analysis chart set is then assembled according to the Kanban structure template to obtain an EEG response analysis chart interface, and the EEG response analysis chart interface is then visualized.

8. A closed-loop analysis system for sound stimulus-EEG response, implementing the closed-loop analysis method for sound stimulus-EEG response as described in any one of claims 1 to 7, characterized in that, The system includes a data acquisition module, a data preprocessing module, a data processing module, a data analysis module, and an intelligent visualization module; The data acquisition module acquires real-time EEG data under sound stimulation based on the standardized interface of the EEG acquisition system. The data preprocessing module performs artifact removal and data standardization operations on the real-time EEG data based on the stimulus event identifiers to obtain standard EEG data. The data processing module performs response latency analysis on the standard EEG data based on the trigger time of the sound stimulus and the peak time of the EEG response to obtain response latency data. Response significance analysis was performed on the standard EEG data based on the dynamic response intensity threshold to obtain response significance data; The data analysis module is used to extract abnormal response event sets from the standard EEG data, and to perform context feature modeling and abnormal root cause analysis on the abnormal response event sets based on the response latency data and response saliency data to obtain abnormal root cause data. The intelligent visualization module generates an EEG response analysis graph interface corresponding to the response latency data, response saliency data, and abnormal root cause data based on neural signal analysis logic, and then visualizes the EEG response analysis graph interface.

Images