A method, system and sensing device for brain-computer interface data acquisition and analysis

Abstract

The application provides a brain-computer interface data acquisition and analysis method, system and sensing device, and belongs to the technical field of data processing. Specifically, brain-computer interface data of a target terminal when the target terminal is combined with an associated sleep monitoring device to verify a sleep state is acquired as feature data, deviation of the brain-computer interface data of the target terminal and the feature data and acquisition data of the brain-computer interface data are obtained, when forgetting processing of the brain-computer interface data is required, forgetting management methods of the brain-computer interface data of different target terminals are determined according to model recognition deviation of different target terminals in a target period, and the demand for storage space in a cloud platform is reduced.

Description

Technical Field

[0001] This invention belongs to the field of data processing technology, and in particular relates to a method, system and sensing device for brain-computer interface data acquisition and analysis. Background Technology

[0002] To achieve monitoring and processing of sleep states during sleep, the invention patent application CN202511003857.3, "A Sleep Classification Method and System Based on Time-Frequency Joint Analysis," proposes efficient modeling and accurate classification of multi-scale EEG signals. While improving deep sleep recognition capabilities and overall classification accuracy, it significantly reduces the number of model parameters and inference latency, exhibiting good real-time performance and clinical applicability. However, it suffers from the following technical problems:

[0003] When performing sleep staging on a cloud platform for users with monitoring devices, differences in user devices or wearing habits may prevent the simultaneous use of heart rate, heart rate variability, activity monitoring signals from body activity monitoring devices, and EEG signals for sleep staging identification. This could lead to difficulties in reliably verifying the accuracy of sleep staging results using EEG signals. Therefore, it is necessary to acquire brain-computer interface (BCI) data from the cloud platform for some users, even when they are not using multiple devices simultaneously. This ensures sufficient data to determine the reliability of sleep analysis using EEG signals during periods of simultaneous use. However, determining a forgetting management strategy for BCI data from monitoring devices to reduce storage pressure while ensuring the reliability of sleep staging results verification becomes a pressing technical problem.

[0004] Therefore, there is an urgent need for a method, system, and sensing device for brain-computer interface data acquisition and analysis. Summary of the Invention

[0005] To achieve the objectives of this invention, the following technical solution is adopted:

[0006] Specifically, this application provides a method for brain-computer interface data acquisition and analysis, which includes:

[0007] S1 uses the data from the EEG signal sensing device to determine the model verification data for all target terminals' brain-computer interface data. Based on the model verification data, when it is determined that there is no need to perform data analysis and processing on all target types of brain-computer interface data on the cloud platform, proceed to the next step.

[0008] S2 determines the model recognition deviation in different target terminals, and combines the usage data of the associated sleep monitoring device of the target terminal to determine the target terminal for data analysis and processing of the target type of brain-computer interface data, and uses it as the target terminal for acquisition;

[0009] S3 uses the brain-computer interface data of the target terminal during the verification process of sleep state in conjunction with the associated sleep monitoring device as feature data. Based on the deviation between the brain-computer interface data of the target terminal and the feature data, as well as the acquired brain-computer interface data, when it is determined that brain-computer interface data forgetting processing is required, different forgetting management methods for the acquired brain-computer interface data of the target terminal are determined according to the model recognition deviation of different target terminals in the target time period.

[0010] The beneficial effects of this invention are as follows:

[0011] Based on the model recognition deviation in different target terminals and the usage data of the associated sleep monitoring devices of the target terminals, the target terminals for data analysis and processing of brain-computer interface data of the target type are determined. By combining the model recognition deviation, the requirements for acquiring and processing brain-computer interface data are determined based on the needs for model update processing. Furthermore, the target terminals for acquisition are determined from the perspective of the needs for acquiring and processing brain-computer interface data and the frequency of use of associated monitoring devices. This ensures that brain-computer interface data can be effectively acquired and processed in target terminals with high usage frequency of associated sleep monitoring devices, thus guaranteeing the reliability of data verification and processing.

[0012] Based on the model recognition deviation of different target terminals in the target time period, different forgetting management methods for acquiring brain-computer interface data from target terminals are determined. This enables the differentiation of the required reliability of brain-computer interface data storage based on the recognition deviation of target terminals in the target time period, i.e., the period since the last abnormal processing of brain-computer interface data. Thus, while ensuring the reliability of model update processing and storage reliability, forgetting processing of brain-computer interface data is performed on some target terminals, reducing storage pressure and ensuring the reliability of model update processing.

[0013] Furthermore, the sensing device is a device for acquiring the user's electroencephalogram (EEG) signals.

[0014] Furthermore, the model verification data of the brain-computer interface data of the target terminal is determined based on the degree of consistency between the EEG signal staging model using brain-computer interface data for sleep staging and the staging model using other signals for sleep staging.

[0015] Furthermore, the other signals include heart rate, heart rate variability, and activity monitoring signals from the body activity monitoring device.

[0016] Furthermore, it was determined that data analysis and processing are not required for all target types of brain-computer interface data on the cloud platform, specifically including:

[0017] Using model validation data from brain-computer interface data of the target terminal, the model validation results of the EEG signal staging model in different target terminals and the staging model using other signals for sleep staging are determined.

[0018] Based on the model validation results, determine the number of model validations on different target terminals;

[0019] The number of model validations on different target terminals determines whether data analysis and processing of all target types of brain-computer interface data is required on the cloud platform.

[0020] Furthermore, the method for determining the forgetting management method for acquiring brain-computer interface data from the target terminal is as follows:

[0021] Based on the model recognition deviation of different target terminals in the target time period, the target terminals with model recognition deviation in the target time period are identified and regarded as the recognition deviation terminals.

[0022] Based on the model recognition data of different target terminals in the target time period, determine the time period during which different target terminals are used simultaneously in the target time period;

[0023] Based on the identified deviation terminal data and the simultaneous use time of different target terminals in the target time period, a forgetting management method for acquiring brain-computer interface data of target terminals is determined.

[0024] In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the aforementioned method for brain-computer interface data acquisition and analysis when running the computer program.

[0025] Thirdly, the present invention provides a sensing device applied to the aforementioned brain-computer interface data acquisition and analysis method, specifically including: electrodes: in direct contact with the scalp to acquire weak electroencephalogram (EEG) signals; electrode caps: an elastic cap with preset electrode positions, wherein the electrodes are fixed in precise positions to ensure consistency between different recordings; and a signal processing module that amplifies the weak signals using an amplifier and converts the continuous analog signals output by the amplifier into discrete digital signals that can be processed by a computer.

[0026] Other features and advantages will be set forth in the following description, and the objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0027] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0028] The above and other features and advantages of the present invention will become more apparent from a detailed description of exemplary embodiments thereof with reference to the accompanying drawings.

[0029] Figure 1 This is a flowchart of a method for data acquisition and analysis in brain-computer interfaces;

[0030] Figure 2 This is a flowchart that determines that data analysis and processing of all target types of brain-computer interface data is unnecessary on the cloud platform;

[0031] Figure 3 This is a flowchart of the method for determining the target terminal. Detailed Implementation

[0032] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0033] Example 1

[0034] like Figure 1 As shown, this application provides a method for brain-computer interface data acquisition and analysis, specifically including:

[0035] S1 uses the data from the EEG signal sensing device to determine the model verification data for all target terminals' brain-computer interface data. Based on the model verification data, when it is determined that there is no need to perform data analysis and processing on all target types of brain-computer interface data on the cloud platform, proceed to the next step.

[0036] Furthermore, the sensing device is a device for acquiring the user's electroencephalogram (EEG) signals.

[0037] Furthermore, the model verification data of the brain-computer interface data of the target terminal is determined based on the degree of consistency between the EEG signal staging model using brain-computer interface data for sleep staging and the staging model using other signals for sleep staging.

[0038] Furthermore, the other signals include heart rate, heart rate variability, and activity monitoring signals from the body activity monitoring device.

[0039] The fundamental purpose of this method is to reduce the unnecessary and massive data processing and storage pressure on cloud platforms. Its core idea is that if there is enough evidence to show that a simplified sleep staging model based on EEG signals is reliable, then the cloud does not need to perform storage and analysis processing when it is not using multiple monitoring devices to analyze and process sleep states simultaneously, thereby reducing the data storage and analysis pressure on the cloud platform.

[0040] Specifically, such as Figure 2 As shown, it is determined that data analysis and processing are not required for all target types of brain-computer interface data on the cloud platform, specifically including:

[0041] The target type of brain-computer interface data refers to data from specific types of terminals that simultaneously collect two types of signals: electroencephalogram (EEG) monitoring signals and other signals such as heart rate, heart rate variability, and body movement. These signals are more easily collected by monitoring devices (such as watches and wristbands).

[0042] EEG signal staging model: A sleep staging model based on EEG signals. Staging model using other signals for sleep staging: A sleep staging model based on alternative signals such as heart rate and body movement.

[0043] S11 uses model validation data from the brain-computer interface data of the target terminal to determine the model validation status of the EEG signal staging model in different target terminals and the staging model that uses other signals for sleep staging processing.

[0044] Model Validation Results: During the same user's sleep on the same night, the staging results of the "alternative model" were compared with those of the "EEG model" to assess their accuracy. Model Validation Numbers: The total number of successful comparative validations for a specific target terminal model. A higher number of validations indicates a more thorough reliability assessment of the "EEG model" on that terminal.

[0045] In the above steps, model validation data is collected, and the "EEG model" and "alternative model" are run synchronously on each target terminal. The sleep stage results (such as deep sleep, light sleep, REM duration, etc.) output by them are compared. The output is: a consistency report of the two model results on each terminal (such as accuracy and Kappa coefficient).

[0046] S12 determines the number of model verifications in different target terminals based on the model verification results;

[0047] Specifically, the number of model verifications is counted. For each target terminal, the total number of successful model comparison verifications is counted, and the output is a list recording [Terminal Model - Number of Verifications].

[0048] S13 determines whether data analysis and processing of all target types of brain-computer interface data is required on the cloud platform based on the number of model validations in different target terminals.

[0049] In the above steps, based on the number of verifications, a cloud processing strategy is determined to identify reliable terminals: Terminals with a model verification count greater than a preset verification count threshold are selected. The "EEG signal staging model" in these target terminals has been thoroughly tested, and the prevalence of reliable terminals is assessed: The number of these "identified reliable terminals" among all target terminals is counted.

[0050] Final decision: If the number of reliable terminals identified is greater than the preset threshold, it is not necessary to perform data analysis and processing on all target types of brain-computer interface data on the cloud platform. There are already enough samples to prove that the "EEG signal staging model" performs stably and reliably on these terminals. Therefore, there is no need to store and analyze the EEG signal data of target terminals that are not used simultaneously with EEG signal monitoring devices and associated monitoring devices.

[0051] Otherwise (if the number of reliable terminals is insufficient), data analysis and processing of brain-computer interface data of all target types need to be performed in the cloud. At this time, the reliability of the "EEG signal staging model" has not been fully verified. The cloud needs to retain all the original data so that when new periods of simultaneous use of EEG signal monitoring equipment and associated monitoring equipment appear, there will be enough data for reference, so as to conduct manual review or model retraining.

[0052] Specifically, the number of times the EEG signal staging model is validated against the staging model that uses other signals for sleep staging is taken as the model validation count.

[0053] Specifically, the target type of brain-computer interface data refers to brain-computer interface data that is simultaneously configured with a target terminal that monitors and processes both EEG signals and other signals.

[0054] Understandably, the number of model validations on different target terminals is used to determine whether data analysis and processing of all target types of brain-computer interface data needs to be performed on the cloud platform, specifically including:

[0055] Based on the number of model validations, target terminals that meet the requirements for the number of model validations are determined (i.e., target terminals whose number of model validations is greater than the preset number of validations threshold).

[0056] Based on the number of target terminals that meet the model validation requirements, determine whether data analysis and processing of all target types of brain-computer interface data is required on the cloud platform.

[0057] It is understandable that when the number of target terminals that meet the model verification requirements is greater than the preset terminal number threshold, the reliability of the EEG signal staging model has been reliably verified. Therefore, it is determined that data analysis and processing of all target types of brain-computer interface data is not required on the cloud platform at this time.

[0058] In one possible embodiment, the first step is to identify a reliable terminal. If the number of verifications (30) of the target terminal is greater than the preset number of verifications threshold (20), the target terminal is marked as a "target terminal whose model verification count meets the requirements".

[0059] Step 2: Assess the availability of reliable terminals;

[0060] Currently, the number of terminals marked as "reliable" is 500. If 500 > the preset terminal number threshold (400), then the cloud platform does not need to perform data analysis and processing on all target types of brain-computer interface data.

[0061] It should be noted that when data analysis and processing of all target types of brain-computer interface data is required, all target types of brain-computer interface data need to be stored. When a user simultaneously uses brain-computer interface data for sleep staging and uses other signals for sleep staging, then all brain-computer interface data in the target terminal are analyzed and processed in a targeted manner. That is, it is determined whether the brain-computer interface data when the user simultaneously uses brain-computer interface data for sleep staging is similar to the brain-computer interface data when using other signals for sleep staging, and whether the staging results of the brain-computer interface data under similar brain-computer interface data are reliable.

[0062] S2 determines the model recognition deviation in different target terminals, and combines the usage data of the associated sleep monitoring device of the target terminal to determine the target terminal for data analysis and processing of the target type of brain-computer interface data, and uses it as the target terminal for acquisition;

[0063] Furthermore, the model recognition deviation in the target terminal is determined based on the degree of consistency between the EEG signal staging model that uses brain-computer interface data for sleep staging and the staging model that uses other signals for sleep staging. For example, if the activity monitoring signal of a body activity monitoring device is used to determine that the user is in an active state, but the staging result of the EEG signal staging model is a sleep state, then it is determined that there is a verification deviation.

[0064] Specifically, the associated monitoring equipment includes heart rate monitoring equipment and body activity monitoring devices, wherein the body activity monitoring devices include an accelerometer, a gyroscope, a magnetometer, and a pressure / vibration sensor installed on the body.

[0065] Specifically, such as Figure 3 As shown, the method for determining the target terminal is as follows:

[0066] The core objective of this process is to accurately identify the terminals that must upload all their brain-computer interface data to the cloud for storage (i.e., "acquiring target terminals"). The ultimate goal is to ensure that there is a sufficient amount of high-quality synchronous data for continuous validation and optimization of EEG signal staging models, preventing the inability to determine model reliability due to insufficient validation data.

[0067] Target terminal model: refers to a specific product model. It is divided into two categories based on hardware configuration:

[0068] Full configuration model: This model of terminal is equipped with built-in or standard EEG signal monitoring equipment and related monitoring equipment (such as heart rate sensor, accelerometer).

[0069] Basic configuration model: This model of terminal only has a built-in EEG signal monitoring device and no associated monitoring devices.

[0070] S21 determines the number of model recognition deviations in different target terminals based on the model recognition deviation in the target terminal;

[0071] Model identification bias: This refers to the discrepancy between the results obtained by the sleep staging model based on EEG signals (the gold standard) and the sleep staging model based on other signals (such as heart rate and body movement) within the same time period (alternative models). For example, an EEG signal showing a high heart rate and frequent body movement indicates that the user is awake, but the EEG model incorrectly identifies the user as being in deep sleep.

[0072] Model identification deviation count: The total number of model identification deviation events that have occurred historically for a specific target terminal. This is the first step in quantifying the reliability of a single terminal. The number of deviations is a direct indicator of the stability of the model's performance on that terminal. The higher the number, the greater the performance problem of the data or model on that terminal.

[0073] The target terminal TB-001 conducted 1000 sleep staging analyses over the past three months. Comparison revealed that in 45 of these analyses, the EEG signal staging model showed inconsistencies with other models. Result: TB-001's model identification error count = 45.

[0074] S22 determines the time period during which the associated sleep monitoring device and the EEG signal sensor of the target terminal are used simultaneously based on the usage data of the associated sleep monitoring device of the target terminal, and takes this as the time period during which they are used simultaneously.

[0075] Simultaneous usage period: Specifically refers to the effective, single time period during which this terminal and its associated device are paired and used.

[0076] Significance of the steps: To statistically analyze the frequency of joint use of this terminal in history and the amount of data that can be provided for reference.

[0077] Device_ZhangSan_001 used both the EEG headband and smartwatch for sleep monitoring on 25 nights in the past month. The number of simultaneous usage periods for Device_ZhangSan_001 in the past month = 25.

[0078] S23 determines the target terminal among the target terminals based on the number of model identification deviations in different target terminals and the data of the time period used simultaneously.

[0079] Specifically, the number of model identification deviations refers to the number of times in the target terminal, the analysis results of the EEG signal staging model are inconsistent with those of staging models that use other signals for sleep staging.

[0080] It is understood that the target terminal is determined based on the number of model identification deviations in different target terminals and the data used during the same time period, specifically including:

[0081] S231 obtains the number of model recognition deviations in different target terminals, and determines whether the number of model recognition deviations in different target terminals is greater than a preset deviation threshold. If so, the target terminals that simultaneously exist in the sensing device and the associated sleep monitoring device are all taken as the target terminals for acquisition. If not, proceed to the next step.

[0082] S231: First-level filtering - Based on the absolute deviation of multiple terminals, a preset deviation number threshold is set as an absolute number threshold.

[0083] If the historical deviation counts of all terminals are very high, it indicates that the staging reliability of the EEG signal staging model may be poor. Therefore, close monitoring is required, that is, all target terminals that simultaneously possess the aforementioned sensing device and the associated sleep monitoring device should be considered as target terminals for acquisition.

[0084] Example: The preset deviation number threshold is set to 15 times. Determine: the average number of model recognition deviations for all target terminals is > 15. If the average number of model recognition deviations for all target terminals is not > 15, proceed to the next step.

[0085] S232 determines the proportion of model recognition deviations in the number of model verifications in different target terminals based on the number of model recognition deviations in different target terminals, and uses this as the deviation proportion. It then determines whether there are target terminals with a deviation proportion greater than a preset deviation proportion threshold. If so, all target terminals that simultaneously exist in the sensing device and the associated sleep monitoring device are considered as acquisition target terminals. If not, proceed to the next step.

[0086] In the above steps, the second layer of screening is based on the relative deviation frequency of certain terminals.

[0087] Definition of terms: Deviation ratio: For this terminal, the proportion of its deviation count to the total number of "simultaneous use periods". Deviation ratio = Number of deviations identified by the model / Total number of simultaneous use periods. Preset deviation ratio threshold: A set ratio threshold.

[0088] If there are terminals with an abnormally high "error frequency", it indicates that the reliability of the EEG signal staging model may be poor. Therefore, close monitoring is required. In this case, all target terminals that simultaneously have the aforementioned sensing device and the associated sleep monitoring device should be considered as target terminals for acquisition.

[0089] Example: The total number of simultaneous usage periods for terminal Device_ZhangSan_001 is 150, with 8 deviations. Calculation: Deviation percentage = 8 / 150 ≈ 0.053 (5.3%). The preset deviation percentage threshold is set to 0.1 (10%). At this time, the deviation percentage of all target terminals is not greater than 0.1, so the determination of target terminals is not made based on this condition, and proceeds to S233.

[0090] S233 determines the recognition deviation probability of the EEG signal staging model based on the average deviation ratio in different target terminals, and judges whether the recognition deviation probability of the EEG signal staging model is less than the preset deviation probability threshold. If yes, proceed to the next step; otherwise, all target terminals that simultaneously exist in the sensing device and the associated sleep monitoring device are regarded as acquisition target terminals among the target terminals.

[0091] S233: Third-level screening - Based on the impact of the global model health on this terminal, the recognition deviation probability of the EEG signal staging model: the deviation ratio of the entire system (total number of deviations across all terminals / total time period of simultaneous use of all terminals), preset deviation probability threshold: the threshold for judging whether the entire system model is healthy and stable.

[0092] If the global model is unhealthy, it means that the EEG signal staging model may be unreliable. In this case, any terminal that can provide verification data (i.e., all terminals that have been used simultaneously with the associated devices) becomes extremely valuable and should all be included in the monitoring to collect data to repair the model.

[0093] Example: Assuming the global deviation probability of the system is 0.042 (4.2%), and the preset deviation probability threshold is set to 0.08 (8%), determine: 0.042 < 0.08? Yes.

[0094] Decision: The global model is healthy and does not require monitoring due to global issues. Proceed to S234.

[0095] S234 Based on the simultaneous use time data in different target terminals, determine whether the number of simultaneous use time periods in the most recent preset time period for different target terminals is greater than the preset use time period number threshold. If yes, proceed to the next step; otherwise, all target terminals that simultaneously exist in the sensing device and the associated sleep monitoring device are regarded as acquisition target terminals among the target terminals.

[0096] S234: Fourth-level filtering - based on the frequency of use during the simultaneous use of all target terminals. Explanation of terms: Preset usage time period threshold: the basic threshold for considering a terminal as having potential value.

[0097] If the number of simultaneous usage periods of all target terminals in the past month is relatively small, it may not be possible to guarantee the reliability and timeliness of data acquisition by acquiring EEG signals from individual target terminals. Therefore, it is necessary to include all target terminals that simultaneously possess the aforementioned sensing device and the associated sleep monitoring device as target terminals for acquisition.

[0098] Example: The number of simultaneous usage periods of all target terminals in the most recent period is 5. The preset threshold for the number of usage periods is set to 2. Judgment: 5 > 1. Decision: At this time, it is not necessary to include all target terminals that simultaneously have the sensor device and the associated sleep monitoring device as target terminals among the target terminals.

[0099] S235 determines whether the number of simultaneous usage periods of the target terminal within the most recent preset time period is greater than the second preset time period number threshold (the second usage period number threshold is greater than the preset usage period number threshold). If yes, the target terminal is designated as the acquisition target terminal; otherwise, the target terminal is determined not to be an acquisition target terminal.

[0100] S235: Fifth Layer Filtering - Core Value of Data Contribution Based on the Terminal. Explanation: Second Preset Time Period Quantity Threshold: A higher threshold used to identify core contributor terminals that can "continuously and stably provide a large amount of high-quality verification data".

[0101] We precisely target high-quality user terminals with excellent habits, using two devices simultaneously almost daily. These terminals form the cornerstone of the verification database; ensuring their data is stored intact guarantees the sustainability of the verification process.

[0102] Example: The monthly simultaneous usage period of terminal Device_ZhangSan_001 is 25 (equivalent to being used synchronously for 25 days in a month). The second preset time period number threshold is set to 20. Final decision: Terminal Device_ZhangSan_001 is determined as the "target terminal for acquisition".

[0103] It should be noted that when the target terminal is a target acquisition terminal, it is determined that all brain-computer interface data of the target terminal needs to be stored and processed.

[0104] S3 uses the brain-computer interface data of the target terminal during the verification process of sleep state in conjunction with the associated sleep monitoring device as feature data. Based on the deviation between the brain-computer interface data of the target terminal and the feature data, as well as the acquired brain-computer interface data, when it is determined that brain-computer interface data forgetting processing is required, different forgetting management methods for the acquired brain-computer interface data of the target terminal are determined according to the model recognition deviation of different target terminals in the target time period.

[0105] It should be noted that the feature data refers to the brain-computer interface signal features of brain-computer interface data during the simultaneous use period, specifically including frequency features and amplitude features.

[0106] Feature data: Features extracted from brain-computer interface data during the "simultaneous use period" that represent reliable EEG patterns. These mainly include: frequency features, such as power spectral density of different frequency bands (δ, θ, α, β, γ), and amplitude features, such as peak value, mean, and variance of EEG waves. These features are considered as "standard templates" for the terminal under known reliable conditions.

[0107] Specifically, the data from "acquiring the target terminal" is uploaded to the cloud in its entirety, but storage space and computing resources are not unlimited. This process aims to determine when it is safe to delete (forget) the historical data of a "acquiring the target terminal." Its core logic is: if a terminal consistently and stably provides high-quality verification data, its history is retained; if the verification data it provides becomes sparse or interrupted, the value of its historical data will decrease over time and can be forgotten.

[0108] Specifically, determining the need for forgetting processing of brain-computer interface data includes:

[0109] S31 uses the acquired brain-computer interface data of the target terminal to determine the deviation between the acquired brain-computer interface data of the target terminal and the feature data, and based on the deviation, determines the deviated interface data in the acquired brain-computer interface data of the target terminal.

[0110] Deviation interface data: refers to brain-computer interface data that differs significantly from the above "standard template". Forgotten processing: deletes historical brain-computer interface data of a specific terminal from cloud storage.

[0111] Bias Assessment: This involves comparing the features of a segment of EEG data with a "feature data" template and calculating the degree of difference. Bias Rate: This is a quantified value of the difference calculated using the Euclidean distance function. The larger the distance, the higher the bias rate, indicating a greater difference between the data segment and the reliable template. Preset Bias Rate Threshold: This is the critical value used to determine whether the data is "abnormal."

[0112] Deviation Interface Data: Brain-computer interface data segments with a deviation rate exceeding a preset deviation rate threshold. This step is a data quality screening. It identifies "low-quality" or "abnormal" data that may be caused by signal interference, device malfunction, or improper wear. This data has little value for model validation.

[0113] Terminal A has been marked as "Target Terminal Acquisition". The system will extract features from the EEG data during the simultaneous use period, calculate the Euclidean distance with the previously accumulated "feature data" template, and calculate the minimum deviation rate of the feature data from the simultaneous use period to be 0.15. The preset deviation rate threshold is 0.1. Judgment: 0.15 > 0.1. Result: This data segment is marked as deviation interface data.

[0114] S32 determines the amount of brain-computer interface data acquired from the target terminal based on the acquired data of the brain-computer interface data of the target terminal.

[0115] Data Acquisition Volume: This refers to the total amount of brain-computer interface data (typically measured in MB or hours) that the target terminal has uploaded to the cloud and stored. It quantifies the storage resources occupied by the terminal in the cloud. This is one of the prerequisites for determining whether "cleaning" is necessary. If the data volume is small, there is no need to forget it.

[0116] Example: Terminal A has stored 200MB of EEG data in the cloud. Result: Terminal A acquires 200MB of data.

[0117] S33 determines whether the target terminal needs to perform forgetting processing on the brain-computer interface data based on the deviation interface data in the brain-computer interface data of the target terminal and the amount of brain-computer interface data acquired.

[0118] It is understood that the deviation interface data is brain-computer interface data whose deviation rate from the frequency feature or amplitude feature of the feature data is greater than a preset deviation rate threshold, and the specific deviation rate is determined according to the Euclidean distance function.

[0119] Furthermore, based on the deviation interface data in the brain-computer interface data of the target terminal and the amount of brain-computer interface data acquired from the target terminal, it is determined whether the target terminal needs to perform forgetting processing of the brain-computer interface data, specifically including:

[0120] S331 determines whether there is biased interface data in the brain-computer interface data of the target terminal. If yes, proceed to the next step. If no, determine that the target terminal does not need to perform forgetting processing on the brain-computer interface data.

[0121] S331: First-level screening - Does low-quality data exist? If there is no low-quality deviation data in the historical data of this terminal, it means that its data quality is consistently high and very valuable, and should not be forgotten. Cleaning is only considered when there is no low-quality data.

[0122] Example: There are marked "deviation interface data" in the historical data of terminal A. Decision: The condition is met, proceed to S332.

[0123] S332 determines whether the amount of brain-computer interface data acquired by the target terminal is greater than a preset data amount threshold based on the amount of data acquired. If yes, proceed to the next step; otherwise, determine that the target terminal does not need to perform forgetting processing on the brain-computer interface data.

[0124] S332: Second-level filtering - Is the data volume large enough? If the data volume is small, even if there is some low-quality data, the storage space benefit of cleaning it is small and the operation is not worthwhile. Cleaning is only meaningful when the data volume is large.

[0125] Example: The preset data volume threshold is set to 100MB. The data volume of terminal A (450 MB) is >100 MB. Decision: The condition is met, proceed to S333.

[0126] S333 Based on the acquired data of the characteristic data of the acquired target terminal, determine the simultaneous use period of the acquired target terminal, and determine whether the acquired target terminal needs to perform forgetting processing of brain-computer interface data based on the simultaneous use period of the acquired target terminal.

[0127] Simultaneous usage period: Specifically refers to the period during which the terminal generates "feature data" (i.e., verifiable data). Preset time period: For example, "the most recent month". Interval between adjacent simultaneous usage periods: Within the preset time period, the time difference between two consecutive "simultaneous usage periods" (for example, the last time was November 5th, and the next time is November 8th, with an interval of 3 days). Preset duration threshold: Defines a threshold for "whether the data stream is continuous", for example, 7 days.

[0128] Significance of this step: This is the most critical decision-making step. It determines whether users are consistently and actively providing high-quality validation data. If the intervals are all less than the threshold, it indicates that user habits are stable and they are continuously providing new validation data. Old data (including biased data) can be replaced by new data, and the system's model validation is continuous, so old data can be safely discarded. If there are intervals longer than the threshold, it indicates that the user's validation data stream has been interrupted. The system cannot continuously validate the model with new data. In this case, retaining historical data (even if partially biased) becomes more important for long-term analysis and model maintenance, and therefore should not be discarded.

[0129] Example: Examine the "simultaneous usage period" of terminal A in the most recent month (preset time period), with the date sequence as: ...[October 25, October 28, November 2, November 20]... Calculate the adjacent intervals: the interval from October 28 to November 2 is approximately 5 days, and the interval from November 2 to November 20 is approximately 18 days. Set the preset duration threshold to 7 days. Determine: Are all intervals less than 7 days? No (because 18 days > 7 days). Final decision: Terminal A needs to undergo forgetting processing of brain-computer interface data.

[0130] Specifically, based on the simultaneous usage period of the target terminal, determining whether the target terminal needs to undergo forgetting processing of brain-computer interface data includes:

[0131] Based on the simultaneous use period of the target terminal, determine the interval between adjacent simultaneous use periods within the most recent preset time period, and determine whether the interval between adjacent simultaneous use periods is less than a preset time threshold. If so, determine that the target terminal does not need to perform forgetting processing of brain-computer interface data; otherwise, determine that the target terminal needs to perform forgetting processing of brain-computer interface data.

[0132] Specifically, the target time period is the period following the last abnormal processing of the brain-computer interface data acquired from the target terminal.

[0133] Target period: This refers to the time from the last time a data forgetting operation was performed on any "acquire target terminal" until the present. This is the observation window for evaluating system stability.

[0134] Terminal with identification deviation: The "Target Terminal" that has experienced "Model Identification Deviation" at least once within the target time period.

[0135] Model identification bias: Within a "simultaneous use period", the results of the EEG signal staging model are inconsistent with those of staging models based on other signals (such as heart rate).

[0136] Forgetting management approach: The final decision at the system level regarding data forgetting, namely, "allowing forgetting" or "prohibiting forgetting".

[0137] Specifically, the method for determining the forgetting management method for acquiring brain-computer interface data from the target terminal is as follows:

[0138] S41 uses the model recognition deviation of different target terminals in the target time period to determine the target terminals that have model recognition deviation in the target time period, and uses them as the recognition deviation terminals.

[0139] S42 determines the simultaneous usage period of different target terminals within the target time period based on the model recognition data of different target terminals within the target time period;

[0140] The statistics are compiled from data of all 1,000 target terminals within the target time period.

[0141] Output: List of terminals exhibiting identification bias: 30 terminals were found to have model identification bias during the target time period. These are the terminals with identification bias. Data on simultaneous usage periods for each terminal: Records the list of "simultaneous usage periods" for each terminal during the target time period and their total duration.

[0142] S43 determines the forgetting management method for acquiring brain-computer interface data of target terminals based on the identified deviation terminal data and the simultaneous use time of different target terminals in the target time period.

[0143] Furthermore, based on the identified deviation terminal data and the simultaneous use periods of different target terminals within the target time period, the forgetting management method for acquiring brain-computer interface data of the target terminal is determined, specifically including:

[0144] S431 determines whether the number of target terminals whose average total duration of simultaneous use is greater than a preset duration threshold is greater than a preset value for the number of target terminals based on the simultaneous use time of different target terminals in the target time period. If yes, proceed to the next step; otherwise, determine that the brain-computer interface data of all target terminals cannot be forgotten.

[0145] First-level screening - Is the overall system data contribution sufficient? Check if most terminals are still actively contributing high-quality validation data. If most terminals become inactive, it indicates that the system is "deactivated." At this point, it is crucial to retain all historical data and not forget it.

[0146] Example: Calculate the average total usage time of 1000 terminals during the same period. Assuming the result is 25 hours / month, determine: 25 > 20 (preset duration threshold)? Yes. Count how many terminals exceed this average? Assuming there are 850 terminals, determine: 850 > 800 (target terminal number preset value)? Yes.

[0147] Decision: System activity is healthy, proceed to S432.

[0148] S432 Based on the identification deviation terminal data, determine whether there is an identification deviation terminal in the target time period. If yes, proceed to the next step. If no, determine all target terminals that need to perform brain-computer interface data forgetting processing and perform brain-computer interface data forgetting processing, that is, perform deviation interface data deletion processing in the cloud.

[0149] The second layer of screening is performed in the above steps - whether there are any unstable signals in the system. If no identification deviation occurs in the system throughout the entire target period, it means that everything is calm and it is safe to clean up.

[0150] Example: Determination: Are there any terminals with identification deviations? Yes (30 units), Decision: The system has unstable signals and cannot be simply allowed to proceed. Proceed to S433.

[0151] S433 obtains the number of the identification deviation terminals and determines whether the number of the identification deviation terminals is greater than the preset deviation terminal number threshold. If yes, it is determined that the brain-computer interface data of all target terminals cannot be forgotten. If no, proceed to the next step.

[0152] S433: Third-level screening - Is there too many unstable terminals? If the number of problematic terminals is large, it may indicate a general system risk (such as model version bugs or general hardware problems). It is essential to prevent forgetting to retain data for investigation.

[0153] Example: Number of identified deviation terminals = 30. Judgment: 30 > 50 (preset threshold for number of deviation terminals)? No. Decision: Problematic terminals are within a controllable range. Proceed to S434.

[0154] S434 Based on the simultaneous use of different identification deviation terminals during the target time period, and the proportion of the number of identification deviation terminals used simultaneously during the target time period, determine the identification deviation factor of different identification deviation terminals, and determine whether there are identification deviation terminals whose identification deviation factor is greater than the preset deviation factor threshold. If so, determine that the brain-computer interface data of all target terminals cannot be forgotten. If not, proceed to the next step.

[0155] S434: Fourth Layer Filtering - Identifying Individual Terminals with "Severe Problems" and Deviance Factors: For a single terminal exhibiting deviation, this is the ratio of the number of "simultaneous usage periods" during the target time period to the total number of "simultaneous usage periods." This measures the frequency of the terminal's problems, identifying terminals that "fail almost every verification." The existence of such terminals may indicate serious hardware failure or extremely rare user cases, requiring the retention of all data for in-depth analysis.

[0156] Example: The recognition bias factors of 30 terminals with recognition bias were examined. Terminal Faulty_Device_05 was found to have a bias factor as high as 0.8 (it failed in 8 out of 10 simultaneous verifications). The question is: Are there any terminals with a bias factor > 0.5? Yes. Decision: There are terminal with serious problems, posing a high risk. It is determined that the brain-computer interface data from all target terminals cannot be processed for forgetting.

[0157] S435 determines the recognition deviation coefficient based on the recognition deviation factors of different recognition deviation terminals and the proportion of the recognition deviation terminals in the target terminals. It then determines whether the recognition deviation coefficient is greater than a preset deviation coefficient threshold. If so, it determines that the brain-computer interface data of all target terminals cannot be forgotten. If not, it proceeds to the next step.

[0158] If no terminals with high deviation factors are found in S434 (e.g., the factor of all deviation terminals is <0.2), the process continues: S435: Fifth layer screening - overall system instability coefficient, identification deviation coefficient: a comprehensive indicator that takes into account both the "prevalence of problematic terminals" and the "severity of their problems". The formula is: ((number of identified deviation terminals / total number of target terminals) + (average identification deviation factor)) / 2. Example: Identification deviation coefficient = (30 / 1000) + 0.2 = 0.115. Judgment: 0.115 is not greater than 0.2, the overall system instability coefficient is very low, proceed to S436.

[0159] S436 determines whether there is a recognition deviation in the simultaneous use period of the target terminal that needs to perform forgetting processing on the brain-computer interface data during the target time period. If yes, it is determined that the brain-computer interface data of the target terminal cannot be forgotten. If no, it is determined that the brain-computer interface data of the target terminal can be forgotten.

[0160] S436: Final micro-level confirmation: After the system-level approval, a final confirmation is made on the 50 terminals that were forgotten in this round of applications: whether they themselves were "innocent" during the target period.

[0161] Example: Examining these 50 terminals, it was found that Device_X had one record of recognition deviation during the target time period. Decision: For Device_X: Prohibit forgetting its data; For the other 49 terminals: Allow the execution of forgetting processing of brain-computer interface data.

[0162] Specifically, the identification deviation during the simultaneous use period refers to the fact that the analysis results of the EEG signal staging model and the staging model that uses other signals for sleep staging are inconsistent during the simultaneous use period.

[0163] Example 2

[0164] In a second aspect, the present invention provides a computer system comprising: a memory and a processor connected in communication, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the aforementioned method for brain-computer interface data acquisition and analysis when running the computer program.

[0165] Example 3

[0166] Thirdly, the present invention provides a sensing device applied to the aforementioned brain-computer interface data acquisition and analysis method, specifically including: electrodes: in direct contact with the scalp to acquire weak electroencephalogram (EEG) signals; electrode caps: an elastic cap with preset electrode positions, wherein the electrodes are fixed in precise positions to ensure consistency between different recordings; and a signal processing module that amplifies the weak signals using an amplifier and converts the continuous analog signals output by the amplifier into discrete digital signals that can be processed by a computer.

[0167] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0168] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0169] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A method for data acquisition and analysis in brain-computer interfaces, characterized in that, Specifically, it includes: Based on the usage data of the EEG signal sensing device, model verification data for brain-computer interface data of all target terminals is determined. When it is determined that there is no need to perform data analysis and processing on all target types of brain-computer interface data on the cloud platform based on the model verification data, proceed to the next step. The model recognition deviation in different target terminals is determined, and the target terminal for data analysis and processing of brain-computer interface data of the target type is determined by combining the usage data of the associated sleep monitoring device of the target terminal, and this target terminal is used as the acquisition target terminal. The brain-computer interface data of the target terminal during sleep state verification processing in conjunction with associated sleep monitoring devices is used as feature data. Based on the deviation between the brain-computer interface data of the target terminal and the feature data, as well as the acquired brain-computer interface data, when it is determined that brain-computer interface data forgetting processing is required, different forgetting management methods for the acquired brain-computer interface data of the target terminal are determined according to the model recognition deviation of different target terminals in the target time period.

2. The method for brain-computer interface data acquisition and analysis as described in claim 1, characterized in that, The sensing device is a device for acquiring the user's electroencephalogram (EEG) signals.

3. The method for brain-computer interface data acquisition and analysis as described in claim 1, characterized in that, The model validation data of the target terminal's brain-computer interface data is determined based on the degree of consistency between the EEG signal staging model using brain-computer interface data for sleep staging and the staging model using other signals for sleep staging.

4. The method for brain-computer interface data acquisition and analysis as described in claim 3, characterized in that, The other signals include heart rate, heart rate variability, and activity monitoring signals from the body activity monitoring device.

5. The method for brain-computer interface data acquisition and analysis as described in claim 1, characterized in that, It was determined that data analysis and processing of all target types of brain-computer interface data is unnecessary on the cloud platform, specifically including: Using model validation data from brain-computer interface data of the target terminal, the model validation results of the EEG signal staging model in different target terminals and the staging model using other signals for sleep staging are determined. Based on the model validation results, determine the number of model validations on different target terminals; The number of model validations on different target terminals determines whether data analysis and processing of all target types of brain-computer interface data is required on the cloud platform.

6. The method for brain-computer interface data acquisition and analysis as described in claim 5, characterized in that, The number of times the EEG signal staging model is validated against staging models that use other signals for sleep staging is taken as the model validation count.

7. The method for brain-computer interface data acquisition and analysis as described in claim 5, characterized in that, The target type of brain-computer interface data refers to brain-computer interface data that is simultaneously configured with a target terminal that monitors and processes both EEG signals and other signals.

8. The method for brain-computer interface data acquisition and analysis as described in claim 1, characterized in that, The method for determining the forgetting management method for acquiring brain-computer interface data from the target terminal is as follows: Based on the model recognition deviation of different target terminals in the target time period, the target terminals with model recognition deviation in the target time period are identified and regarded as the recognition deviation terminals. Based on the model recognition data of different target terminals in the target time period, determine the time period during which different target terminals are used simultaneously in the target time period; Based on the identified deviation terminal data and the simultaneous use time of different target terminals in the target time period, a forgetting management method for acquiring brain-computer interface data of target terminals is determined.

9. A computer system, comprising: A memory and processor connected in communication, and a computer program stored in the memory and capable of running on the processor, characterized in that, when the processor runs the computer program, it executes a method for brain-computer interface data acquisition and analysis as described in any one of claims 1-8.

10. A sensing device, applied to the brain-computer interface data acquisition and analysis method according to any one of claims 1-8, characterized in that, Specifically, it includes: electrodes that directly contact the scalp to collect weak EEG signals; an electrode cap that is an elastic cap with pre-set electrode positions, which fixes the electrodes in precise positions to ensure consistency between different recordings; and a signal processing module that uses an amplifier to amplify the weak signals and converts the continuous analog signals output by the amplifier into discrete digital signals that can be processed by a computer.

Images