A cognitive regulation method and system based on an interesting task and real-time electroencephalogram signals

Abstract

The application relates to a cognitive regulation method and system based on an interesting task and real-time electroencephalogram signals, wherein a front-end interaction module acquires cognitive regulation requirements of a user; according to the cognitive regulation requirements of the user, a front-end signal analysis and processing module loads corresponding cognitive data sets from a cloud database, carries out pretreatment and feature extraction on the cognitive data sets, and constructs an electroencephalogram pattern recognition classifier by using a machine learning algorithm; the front-end interaction module loads an interesting neural feedback gamified training page; an electroencephalogram acquisition device acquires real-time electroencephalogram signals of the user, and converts the real-time electroencephalogram signals into real-time digital signals; the real-time digital signals are transmitted to the front-end signal analysis and processing module in a Bluetooth communication mode; the front-end signal analysis and processing module carries out pretreatment and feature extraction on the real-time digital signals, obtains characteristic values, and carries out interesting neural feedback gamified training by using the electroencephalogram pattern recognition classifier; after the interesting neural feedback gamified training is completed, the front-end interaction module sends a training report to the user.

Description

Technical Field

[0001] This invention relates to a cognitive modulation method and system based on engaging tasks and real-time EEG signals, belonging to the fields of sensor technology and EEG signal processing technology. Background Technology

[0002] As a precursor to brain-computer interface technology, EEG neurofeedback technology can collect and extract brainwave signal features in real time and provide these features to users in a visual form. Users can then adopt personalized cognitive strategies based on this feedback to self-regulate and thus change their cognitive state. EEG neurofeedback technology can help users improve concentration, relieve stress, and alleviate anxiety in daily life. Based on these promising applications, EEG neurofeedback has developed rapidly and is now gradually entering the commercialization stage.

[0003] Electroencephalography (EEG) neurofeedback uses an EEG cap to collect subtle changes in electrical potential on the scalp, decodes signal features based on machine learning algorithms, and provides real-time feedback primarily in the form of visual input. Currently, the visual feedback in EEG neurofeedback technology is mostly presented as simple bar charts or line graphs, which makes the technology lack sufficient engagement and a user-friendly interface, significantly limiting its commercial application. Furthermore, different cognitive modulations correspond to different EEG signal characteristics, but current EEG neurofeedback technologies lack specificity for cognitive modulation, failing to meet users' needs in various usage scenarios. Summary of the Invention

[0004] This invention addresses the following technical problem: overcoming the shortcomings of existing technologies, it provides a cognitive modulation method and system based on engaging tasks and real-time EEG signals. This solves the problems of EEG neurofeedback technology lacking engaging and user-friendly interfaces, and the signal characteristics decoded by EEG neurofeedback technology lacking specificity for cognitive modulation. These problems significantly limit the commercial application of EEG neurofeedback technology. This invention boasts advantages such as high autonomy, universality, wide applicability, simple operation, strong interactivity, and high level of engagement.

[0005] The technical solution of this invention:

[0006] In a first aspect, the present invention provides a cognitive modulation method based on engaging tasks and real-time EEG signals, comprising the following steps:

[0007] S1, the front-end interaction module obtains the user's cognitive adjustment needs;

[0008] S2, based on the user's cognitive adjustment needs, the front-end signal analysis and processing module loads the corresponding cognitive dataset from the cloud database, preprocesses and extracts features from the cognitive dataset, and uses machine learning algorithms to build an EEG pattern recognition classifier;

[0009] S3, the front-end interactive module loads an engaging neurofeedback gamified training page;

[0010] S4, the EEG acquisition device acquires the user's real-time EEG signals and converts them into real-time digital signals, which are then transmitted to the front-end signal analysis and processing module via Bluetooth communication.

[0011] S5, the front-end signal analysis and processing module preprocesses and extracts features from real-time digital signals to obtain feature values, and uses an EEG pattern recognition classifier to conduct fun neurofeedback gamified training.

[0012] S6: After the gamified neurofeedback training is completed, the front-end interaction module sends a training report to the user.

[0013] Furthermore, in step S1, the cognitive adjustment needs include: 1) relieving life stress; 2) improving anxiety; 3) enhancing concentration; and 4) reducing cravings for smoking.

[0014] Furthermore, S2 is specifically implemented as follows: the cloud database contains different types of cognitive datasets, including: stress-related datasets, anxiety-related datasets, concentration-related datasets, and smoking craving-related datasets, which respectively correspond to the cognitive regulation needs of relieving life stress, improving anxiety, enhancing concentration, and reducing smoking cravings; each type of cognitive dataset includes sub-datasets under two conditions: high threshold and low threshold; after the user selects the type of cognitive regulation need, the front-end signal analysis and processing module loads the corresponding type of cognitive dataset from the cloud database via WIFI communication connection; after loading, the front-end signal analysis and processing module performs preprocessing and feature extraction on the cognitive dataset;

[0015] The preprocessing steps include: high-pass filtering and low-pass filtering of the cognitive dataset; downsampling; average rereference; correction of blink artifacts using the traditional recursive least squares algorithm; segmentation; removal of noisy segments using an automatic detection algorithm; and feature extraction from the preprocessed cognitive dataset.

[0016] The feature extraction steps include: calculating the global field power (GFP) of the cognitive dataset to obtain a global field power time series; inputting the EEG topographic maps corresponding to all peaks in the global field power time series into a k-means clustering algorithm, where k is set to 4, to obtain four standard EEG topographic map categories A, B, C, and D; calculating the spatial correlation coefficient between each peak and the four standard topographic maps, assigning the peak to the category with the highest spatial correlation coefficient, and considering the troughs between peaks as the transition points between different categories; thus, the global field power time series is classified; for different types of cognitive datasets, the extracted feature values ​​include two conditions: high threshold and low threshold.

[0017] For the stress-related dataset, the extracted feature value is: the coverage ratio of class A in the global field power time series. The coverage ratio is defined as the proportion of the total duration of each class to the total duration of the global field power time series.

[0018] The anxiety-related dataset extracts the following features: the transition probability from class A to class D and the transition probability from class B to class C in the global field power time series. The transition probability is defined as the probability of transitioning from one class to another in the global field power time series.

[0019] The attention-related dataset extracted features as follows: the duration and occurrence rate of class D in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series, and the occurrence rate is defined as the number of times each class appears in one second in the global field power time series.

[0020] The extracted feature value from the smoking craving related dataset is the duration of class C in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series.

[0021] The extracted feature values ​​under the two conditions are input into the support vector machine algorithm for classification modeling, thereby constructing an EEG pattern recognition classifier. This EEG pattern recognition classifier can identify and classify a new EEG signal and assign it to different cognitive state conditions.

[0022] Furthermore, S3 is specifically implemented as follows: after the front-end signal analysis and processing module completes the construction of the EEG pattern recognition classifier, the front-end interaction module automatically loads the neurofeedback gamification training page, which includes: 1) game name - Protect Cells; 2) game instructions; 3) start game button; after reading the game instructions, the user clicks the start game button to enter the fun neurofeedback gamification training scenario.

[0023] Furthermore, S4 is specifically implemented as follows: the user wears a smart wearable brain ring device, which uses dry electrodes and a sampling rate of 1000 Hz to collect the user's EEG signals in real time; the signal conversion device amplifies, reduces noise, filters and converts the real-time EEG signals collected by the smart wearable brain ring device into corresponding real-time digital signals; the signal conversion device transmits the real-time digital signals to the front-end signal analysis and processing module via Bluetooth communication.

[0024] Furthermore, step S5 is specifically implemented as follows: the front-end signal analysis and processing module receives real-time digital signals from the signal conversion device at regular intervals via Bluetooth communication, and performs preprocessing and feature extraction on the real-time digital signals;

[0025] The preprocessing steps include: high-pass filtering and low-pass filtering of the real-time digital signal; downsampling; average rereference; correction of blink artifacts using the traditional recursive least squares algorithm; segmentation; removal of noisy segments using an automatic detection algorithm; and feature extraction of the preprocessed real-time digital signal.

[0026] The feature extraction steps include: calculating the global field power (GFP) of the real-time digital signal to obtain a global field power time series; inputting the EEG topography maps corresponding to all peaks in the global field power time series into a k-means clustering algorithm, where k is set to 4, to obtain four standard EEG topography map categories A, B, C, and D; calculating the spatial correlation coefficient between each peak and the four standard topography maps, assigning the peak to the category with the highest spatial correlation coefficient, and considering the troughs between peaks as transition points between different categories; thus, the global field power time series is classified; based on the cognitive modulation needs selected by the user in the initial stage, the extracted feature values ​​are:

[0027] To alleviate life stress, the extracted feature value is: the coverage ratio of class A in the global field power time series. The coverage ratio is defined as the proportion of the total duration of each class to the total duration of the global field power time series.

[0028] To improve anxiety, the extracted features are: the conversion probability from class A to class D and the conversion probability from class B to class C in the global field power time series. The conversion probability is defined as the probability of conversion from one class to another in the global field power time series.

[0029] To enhance focus, the extracted features are: the duration and occurrence rate of class D in the global field power time series. The duration is defined as the average time each class appears in the global field power time series, and the occurrence rate is defined as the number of times each class appears in one second in the global field power time series.

[0030] To reduce cravings for cigarettes, the extracted feature value is the duration of class C in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series.

[0031] The front-end signal analysis and processing module inputs the extracted feature values ​​into the previously constructed EEG pattern recognition classifier to obtain the probability values ​​output by the EEG pattern recognition classifier. Subsequently, the front-end signal analysis and processing module sends the probability values ​​to the front-end interaction module, which performs fun neurofeedback gamification training based on the probability values.

[0032] Furthermore, step S6 is specifically implemented as follows: in the front-end signal analysis and processing module, the average value of all probability values ​​output by the EEG pattern recognition classifier in one training session is calculated, and the average value is sent to the front-end interaction module, which generates a training report based on the average value.

[0033] Secondly, the present invention provides a cognitive regulation system based on fun tasks and real-time EEG signals, including a front end, an EEG acquisition device and a cloud database;

[0034] The front end includes an interaction module and a signal analysis and processing module; the interaction module is used to: 1) obtain the user's cognitive adjustment needs; 2) load an engaging neurofeedback gamified training page; 3) conduct engaging neurofeedback gamified training; and 4) send training reports to the user.

[0035] The signal analysis and processing module is used to: 1) load a cognitive dataset from a cloud database; 2) preprocess and extract features from the cognitive dataset, and construct an EEG pattern recognition classifier using machine learning algorithms; 3) acquire real-time digital signals from an EEG acquisition device; 4) preprocess and extract features from the real-time digital signals, and use the signal features as input values ​​for the EEG pattern recognition classifier to obtain output values.

[0036] The EEG acquisition device includes a smart wearable brain ring device and a signal conversion device. The smart wearable brain ring device contains 32 electrodes, which are dry electrodes. The signal conversion device is integrated into the smart wearable brain ring device. The connection between the signal conversion device and the smart wearable brain ring device is an electrical connection. The connection between the signal conversion device and the front-end signal analysis and processing module is a Bluetooth communication connection.

[0037] The cloud database is used to store different types of cognitive datasets. The front-end signal analysis and processing module loads the aforementioned cognitive datasets from the cloud database. The connection between the cloud database and the front-end signal analysis and processing module is a WIFI communication connection.

[0038] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor.

[0039] The advantages of this invention compared to the prior art are:

[0040] (1) This invention provides a cognitive modulation method and system based on fun tasks and real-time EEG signals. It offers multiple cognitive modulation options through a front-end interactive module; loads corresponding cognitive datasets through a cloud database; constructs an EEG pattern recognition classifier using machine learning algorithms; acquires the user's EEG signals through an EEG acquisition device; and conducts cognitive modulation training through fun neurofeedback games. This greatly solves the problem of the lack of fun and user-friendly human-computer interaction in EEG neurofeedback technology, and can more effectively help users improve concentration, relieve stress, and alleviate anxiety in daily life. Furthermore, it provides a systematic analysis method for various cognitive modulation-related EEG signal characteristics, which can meet the needs of users in different usage scenarios.

[0041] (2) This invention has the advantages of high autonomy, universality, rich applicable scenarios, simple operation, strong interactivity and fun, which can greatly promote the commercialization of EEG neurofeedback technology. Attached Figure Description

[0042] Figure 1 This is a flowchart illustrating the implementation of the method of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0044] like Figure 1 As shown, a cognitive modulation system based on fun tasks and real-time EEG signals is disclosed. The system includes: a front-end, an EEG acquisition device, and a cloud database.

[0045] The front end includes an interaction module and a signal analysis and processing module. The interaction module is used to: 1) obtain the user's cognitive adjustment needs; 2) load an engaging neurofeedback gamified training page; 3) conduct engaging neurofeedback gamified training; and 4) send training reports to the user.

[0046] The signal analysis and processing module is used to: 1) load a cognitive dataset from a cloud database; 2) preprocess and extract features from the cognitive dataset, and construct an EEG pattern recognition classifier using machine learning algorithms; 3) acquire real-time digital signals from an EEG acquisition device; 4) preprocess and extract features from the real-time digital signals, and use the signal features as input values ​​for the EEG pattern recognition classifier to obtain output values.

[0047] The EEG acquisition device includes a smart wearable brain ring device and a signal conversion device. The smart wearable brain ring device contains 32 electrodes (such as those in the Muse and Emotive brain ring devices), and these electrodes are dry electrodes. The signal conversion device is integrated into the smart wearable brain ring device. The connection between the signal conversion device and the smart wearable brain ring device is an electrical connection, and the connection between the signal conversion device and the front-end signal analysis and processing module is a Bluetooth communication connection.

[0048] The cloud database is used to store different types of cognitive datasets. The front-end signal analysis and processing module loads the aforementioned cognitive datasets from the cloud database. The connection between the cloud database and the front-end signal analysis and processing module is a WIFI communication connection.

[0049] First, S1 includes a front-end interaction module that acquires the user's cognitive regulation needs. Specifically, the front-end interaction module provides the user with the following options: 1) relieve life stress; 2) improve anxiety; 3) enhance concentration; 4) reduce cravings for cigarettes. The user can select any one of these cognitive regulation needs by clicking.

[0050] S2 includes, based on the user's cognitive regulation needs, the front-end signal analysis and processing module loading the corresponding cognitive dataset from the cloud database, preprocessing and extracting features from the cognitive dataset, and constructing an EEG pattern recognition classifier using machine learning algorithms. Specifically, the cloud database contains different types of cognitive datasets, including: 1) stress-related datasets, further divided into high-threshold stress-related subsets and low-threshold stress-related subsets; 2) anxiety-related datasets, further divided into high-threshold anxiety-related subsets and low-threshold anxiety-related subsets; 3) attention-related datasets, further divided into high-threshold attention-related subsets and low-threshold attention-related subsets; 4) smoking craving-related datasets, further divided into high-threshold smoking craving-related subsets and low-threshold smoking craving-related subsets. After the user selects a specific type of cognitive regulation need, the front-end signal analysis and processing module loads the corresponding type of cognitive dataset from the cloud database via a Wi-Fi communication connection. Each type of cognitive dataset includes subsets under both high-threshold and low-threshold conditions. After loading, the front-end signal analysis and processing module preprocesses and extracts features from the cognitive dataset.

[0051] The preprocessing steps include: high-pass filtering (2 Hz) and low-pass filtering (20 Hz) of the cognitive dataset; reducing the sampling rate of the cognitive dataset to 250 Hz; performing average rereference; using the traditional recursive least squares algorithm to correct blink artifacts; segmenting the cognitive dataset into segments of 1 second length; using an automatic detection algorithm to remove segments containing noise, with the algorithm's detection criterion being that segments with signal amplitude changes exceeding ±100 mV are considered to contain noise; and then, feature extraction is performed on the preprocessed cognitive dataset.

[0052] The feature extraction steps include: calculating the global field power (GFP) of the cognitive dataset, as shown in the following formula:

[0053] (1)

[0054] In the above formula, N represents the number of electrodes. The potential of electrode i at a given time point. The average potential of all electrodes is represented by the formula, and the result is a global field power time series.

[0055] Next, in the global field power time series, the EEG topographic maps corresponding to all peaks are input into the k-means clustering algorithm, where k is set to 4, resulting in four standard EEG topographic map categories A, B, C, and D. The spatial correlation coefficient between each peak and the four standard topographic maps is calculated, and the peak is assigned to the category with the highest spatial correlation coefficient. The troughs between peaks are considered as the transition points between different categories. At this point, the global field power time series has been classified.

[0056] Next, for different types of cognitive datasets, the subsequently extracted feature values ​​are as follows:

[0057] Cognitive dataset types Extracted feature values Stress-related datasets (high threshold / low threshold subsets) The coverage ratio of Class A in the global field power time series is defined as the proportion of the total duration of each class to the total duration of the global field power time series. Anxiety-related datasets (high-threshold / low-threshold sub-datasets) The transition probability from class A to class D and the transition probability from class B to class C in the global field power time series are defined as the probability of transitioning from one class to another in the global field power time series. Attention-related datasets (high threshold / low threshold sub-datasets) The duration and occurrence rate of class D in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series, and the occurrence rate is defined as the number of times each class occurs in one second in the global field power time series. Smoking craving related datasets (high threshold / low threshold subsets) The duration of class C in the global field power time series is defined as the average time of occurrence of each class in the global field power time series.

[0058] The extracted feature values ​​include both high and low threshold conditions. These feature values ​​are then input into a support vector machine (SVM) algorithm for classification modeling, thereby constructing an EEG pattern recognition classifier. This classifier can identify and classify a new EEG signal into different cognitive state conditions (e.g., high concentration state). The input to the EEG pattern recognition classifier is a new EEG signal feature value, and the output is a probability value ranging from 0 to 1. 0 represents the optimal cognitive state condition (e.g., lowest stress state, lowest anxiety state, highest concentration state, or lowest craving for cigarettes), and 1 represents the worst cognitive state (e.g., highest stress state, highest anxiety state, lowest concentration state, or highest craving for cigarettes). The EEG pattern recognition classifier will be used for subsequent engaging neurofeedback gamification training.

[0059] S3 includes a front-end interaction module that loads an engaging neurofeedback gamified training page. Specifically, after the front-end signal analysis and processing module completes the construction of the EEG pattern recognition classifier, the front-end interaction module automatically loads the neurofeedback gamified training page. The page includes: 1) Game name – Protect Cells; 2) Game instructions; 3) A "Start Game" button. After reading the game instructions, the user clicks the "Start Game" button to enter the engaging neurofeedback gamified training scenario.

[0060] S4 includes an EEG acquisition device that acquires the user's real-time EEG signals and converts them into real-time digital signals, which are then transmitted to the front-end signal analysis and processing module via Bluetooth communication. Specifically, the EEG acquisition device includes a smart wearable brainband device and a signal conversion device. The user wears a smart wearable brainband device (such as the Muse and Emotive brainband devices), which has 32 electrodes configured according to the "International 10-20 System". The device uses dry electrodes and a sampling rate of 1000 Hz to acquire the user's EEG signals in real time. The signal conversion device amplifies, reduces noise, filters, and performs analog-to-digital conversion on the real-time EEG signals acquired by the smart wearable brainband device to obtain the corresponding real-time digital signals. The signal conversion device then transmits the real-time digital signals to the front-end signal analysis and processing module via Bluetooth communication.

[0061] The S5 includes a front-end signal analysis and processing module that preprocesses and extracts features from real-time digital signals to obtain feature values, and uses an EEG pattern recognition classifier for engaging neurofeedback gamified training. Specifically, the front-end signal analysis and processing module receives real-time digital signals from a signal conversion device via Bluetooth communication. The receiving method involves receiving a 5-second segment of real-time digital signal every 5 seconds, wherein the 5-second interval is used for preprocessing and feature extraction of the 5-second real-time digital signal.

[0062] The preprocessing steps include: high-pass filtering (2 Hz) and low-pass filtering (20 Hz) of the real-time digital signal; reducing the sampling rate of the real-time digital signal to 250 Hz; performing average rereference; using the traditional recursive least squares algorithm to correct blink artifacts; segmenting the real-time digital signal into segments of 1 second length; using an automatic detection algorithm to remove segments containing noise, with the algorithm's detection criterion being that segments with signal amplitude changes exceeding ±100 mV are considered to contain noise; and then, feature extraction is performed on the preprocessed real-time digital signal.

[0063] The feature extraction steps include: calculating the global field power (GFP) of the real-time digital signal, with the specific formula as follows:

[0064] (1)

[0065] In the above formula, N represents the number of electrodes. The potential of electrode i at a given time point. The average potential of all electrodes is represented by the formula, and the result is a global field power time series.

[0066] Next, in the global field power time series, the EEG topographic maps corresponding to all peaks are input into the k-means clustering algorithm, where k is set to 4, resulting in four standard EEG topographic map categories A, B, C, and D. The spatial correlation coefficient between each peak and the four standard topographic maps is calculated, and the peak is assigned to the category with the highest spatial correlation coefficient. The troughs between peaks are considered as the transition points between different categories. At this point, the global field power time series has been classified.

[0067] Next, based on the cognitive adjustment needs selected by the user in the initial stage, the subsequently extracted feature values ​​are as follows:

[0068] User-selected cognitive adjustment needs Extracted feature values Relieve life stress The coverage ratio of Class A in the global field power time series is defined as the proportion of the total duration of each class to the total duration of the global field power time series. Improve anxiety The transition probability from class A to class D and the transition probability from class B to class C in the global field power time series are defined as the probability of transitioning from one class to another in the global field power time series. Improve focus The duration and occurrence rate of class D in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series, and the occurrence rate is defined as the number of times each class occurs in one second in the global field power time series. Reduce smoking cravings The duration of class C in the global field power time series is defined as the average time of occurrence of each class in the global field power time series.

[0069] Next, the front-end signal analysis and processing module inputs the extracted feature values ​​into the previously constructed EEG pattern recognition classifier to obtain the probability values ​​output by the EEG pattern recognition classifier (within the range of 0-1). Subsequently, the front-end signal analysis and processing module sends the probability values ​​to the front-end interaction module. Since the front-end signal analysis and processing module receives real-time digital signals for 5 seconds at 5-second intervals, the front-end interaction module receives the probability values ​​every 10 seconds. Based on the probability values, the front-end interaction module performs fun neurofeedback gamification training. The game interface includes a cartoon "cell" and... A cartoon-style "monster" plays a game where the health of "cells" is closely related to a probability value. A probability value closer to 0 indicates a healthier cell, and the cell is redder; a probability value closer to 1 indicates an unhealthier cell, and the cell is blacker. The "monster" attacks "cells" with "viruses" every 5 seconds. When the probability value is less than 0.5, the cell gains a "shield" that protects it from the "viruses" and turns red; when the probability value is greater than 0.5, the cell loses its shield and becomes vulnerable to the "viruses," turning black. In the game, the user's task is to protect the cell's health from the "monster's" attacks. This requires the user to use personalized cognitive strategies to adjust their own state, minimizing the probability value output by the classifier. The underlying principle is that lower probability values ​​represent better cognitive states (e.g., low-stress, low-anxiety, high-attention, or low-craving states), while higher probability values ​​represent worse cognitive states (e.g., maximum-stress, maximum-anxiety, minimum-attention, or maximum-craving states). The engaging neurofeedback gamified training consists of 120 rounds, each lasting 10 seconds, for a total of 20 minutes.

[0070] S6 includes sending a training report to the user after the completion of the gamified neurofeedback training. Specifically, in the front-end signal analysis and processing module, the average value of the 120 probability values ​​output by the EEG pattern recognition classifier in one training session is calculated, and the average value is sent to the front-end interaction module. The front-end interaction module generates a corresponding training report based on the range of the average value: if the average value is in the range of 0-0.25, the training report sent to the user is: "You have perfectly protected the health of your cells, please keep it up!"; if the average value is in the range of 0.25-0.5, the training report sent to the user is: "You have protected the health of your cells quite well! Please keep up the good work!"; if the average value is in the range of 0.5-1, the training report sent to the user is: "You failed to protect the health of your cells, please keep trying!".

[0071] While specific implementation methods of the present invention have been described above, those skilled in the art should understand that these are merely illustrative examples. Various changes or modifications can be made to these implementation methods without departing from the principles and implementation of the present invention. Therefore, the scope of protection of the present invention is defined by the appended claims.

Claims

1. A cognitive modulation method based on engaging tasks and real-time EEG signals, characterized in that, Includes the following steps: S1, the front-end interaction module obtains the user's cognitive adjustment needs; S2, based on the user's cognitive adjustment needs, the front-end signal analysis and processing module loads the corresponding cognitive dataset from the cloud database, preprocesses and extracts features from the cognitive dataset, and uses machine learning algorithms to build an EEG pattern recognition classifier; S3, the front-end interactive module loads an engaging neurofeedback gamified training page; S4, the EEG acquisition device acquires the user's real-time EEG signals and converts them into real-time digital signals, which are then transmitted to the front-end signal analysis and processing module via Bluetooth communication. S5, the front-end signal analysis and processing module preprocesses and extracts features from real-time digital signals to obtain feature values, and uses an EEG pattern recognition classifier to conduct fun neurofeedback gamified training. S6: After the fun neurofeedback gamified training is completed, the front-end interaction module sends a training report to the user. In step S1, the cognitive adjustment needs include: 1) relieving life stress; 2) improving anxiety; 3) enhancing concentration; and 4) reducing cravings for cigarettes. Specifically, S2 is implemented as follows: the cloud database contains different types of cognitive datasets, including: stress-related datasets, anxiety-related datasets, concentration-related datasets, and smoking craving-related datasets, which correspond to the cognitive regulation needs of relieving life stress, improving anxiety, enhancing concentration, and reducing smoking cravings, respectively; each type of cognitive dataset includes sub-datasets under two conditions: high threshold and low threshold; after the user selects the type of cognitive regulation need, the front-end signal analysis and processing module loads the corresponding type of cognitive dataset from the cloud database via WIFI communication connection; after loading, the front-end signal analysis and processing module performs preprocessing and feature extraction on the cognitive dataset; The preprocessing steps include: high-pass filtering and low-pass filtering of the cognitive dataset; downsampling; average rereference; correction of blink artifacts using the traditional recursive least squares algorithm; segmentation; removal of noisy segments using an automatic detection algorithm; and feature extraction from the preprocessed cognitive dataset. The feature extraction steps include: calculating the global field power (GFP) of the cognitive dataset to obtain a global field power time series; inputting the EEG topographic maps corresponding to all peaks in the global field power time series into a k-means clustering algorithm, where k is set to 4, to obtain four standard EEG topographic map categories A, B, C, and D; calculating the spatial correlation coefficient between each peak and the four standard topographic maps, assigning the peak to the category with the highest spatial correlation coefficient, and considering the troughs between peaks as the transition points between different categories; thus, the global field power time series is classified; for different types of cognitive datasets, the extracted feature values ​​include two conditions: high threshold and low threshold. For the stress-related dataset, the extracted feature value is: the coverage ratio of class A in the global field power time series. The coverage ratio is defined as the proportion of the total duration of each class to the total duration of the global field power time series. The anxiety-related dataset extracts the following features: the transition probability from class A to class D and the transition probability from class B to class C in the global field power time series. The transition probability is defined as the probability of transitioning from one class to another in the global field power time series. The attention-related dataset extracted features as follows: the duration and occurrence rate of class D in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series, and the occurrence rate is defined as the number of times each class appears in one second in the global field power time series. The extracted feature value from the smoking craving related dataset is the duration of class C in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series. The extracted feature values ​​under the two conditions are input into the support vector machine algorithm for classification modeling, thereby constructing an EEG pattern recognition classifier; this EEG pattern recognition classifier can identify and classify a new EEG signal and classify it into different cognitive state conditions. The specific implementation of step S5 is as follows: the front-end signal analysis and processing module receives real-time digital signals from the signal conversion device at regular intervals via Bluetooth communication, and performs preprocessing and feature extraction on the real-time digital signals; The preprocessing steps include: high-pass filtering and low-pass filtering of the real-time digital signal; downsampling; average rereference; correction of blink artifacts using the traditional recursive least squares algorithm; segmentation; removal of noisy segments using an automatic detection algorithm; and feature extraction of the preprocessed real-time digital signal. The feature extraction steps include: calculating the global field power (GFP) of the real-time digital signal to obtain a global field power time series; inputting the EEG topography maps corresponding to all peaks in the global field power time series into a k-means clustering algorithm, where k is set to 4, to obtain four standard EEG topography map categories A, B, C, and D; calculating the spatial correlation coefficient between each peak and the four standard topography maps, assigning the peak to the category with the highest spatial correlation coefficient, and considering the troughs between peaks as transition points between different categories; thus, the global field power time series is classified; based on the cognitive modulation needs selected by the user in the initial stage, the extracted feature values ​​are: To alleviate life stress, the extracted feature value is: the coverage ratio of class A in the global field power time series. The coverage ratio is defined as the proportion of the total duration of each class to the total duration of the global field power time series. To improve anxiety, the extracted features are: the conversion probability from class A to class D and the conversion probability from class B to class C in the global field power time series. The conversion probability is defined as the probability of conversion from one class to another in the global field power time series. To enhance focus, the extracted features are: the duration and occurrence rate of class D in the global field power time series. The duration is defined as the average time each class appears in the global field power time series, and the occurrence rate is defined as the number of times each class appears in one second in the global field power time series. To reduce cravings for cigarettes, the extracted feature value is the duration of class C in the global field power time series. The duration is defined as the average time of occurrence of each class in the global field power time series. The front-end signal analysis and processing module inputs the extracted feature values ​​into the previously constructed EEG pattern recognition classifier to obtain the probability values ​​output by the EEG pattern recognition classifier. Subsequently, the front-end signal analysis and processing module sends the probability values ​​to the front-end interaction module, which performs fun neurofeedback gamification training based on the probability values.

2. The cognitive modulation method based on engaging tasks and real-time EEG signals according to claim 1, characterized in that, The S3 is specifically implemented as follows: after the front-end signal analysis and processing module completes the construction of the EEG pattern recognition classifier, the front-end interaction module automatically loads the neurofeedback gamified training page, which includes: 1) game name - Protect Cells; 2) game instructions; 3) start game button; After reading the game instructions, users click the "Start Game" button to enter a fun neurofeedback gamified training scenario.

3. The cognitive modulation method based on engaging tasks and real-time EEG signals according to claim 1, characterized in that, Specifically, S4 is implemented as follows: the user wears a smart wearable brain ring device, which uses dry electrodes and a sampling rate of 1000 Hz to collect the user's EEG signals in real time; the signal conversion device amplifies, reduces noise, filters and converts the real-time EEG signals collected by the smart wearable brain ring device into corresponding real-time digital signals; the signal conversion device transmits the real-time digital signals to the front-end signal analysis and processing module via Bluetooth communication.

4. The cognitive modulation method based on engaging tasks and real-time EEG signals according to claim 1, characterized in that, The specific implementation of step S6 is as follows: In the front-end signal analysis and processing module, the average value of all probability values ​​output by the EEG pattern recognition classifier in one training session is calculated, and the average value is sent to the front-end interaction module. The front-end interaction module generates a training report based on the average value.

5. A cognitive modulation system based on engaging tasks and real-time EEG signals, employing the method described in claim 1, characterized in that: Front-end, EEG acquisition equipment, and cloud database; The front end includes an interaction module and a signal analysis and processing module; the interaction module is used to: 1) obtain the user's cognitive adjustment needs; 2) load an engaging neurofeedback gamified training page; 3) conduct engaging neurofeedback gamified training; and 4) send training reports to the user. The signal analysis and processing module is used to: 1) load a cognitive dataset from a cloud database; 2) preprocess and extract features from the cognitive dataset, and construct an EEG pattern recognition classifier using machine learning algorithms; 3) acquire real-time digital signals from an EEG acquisition device; 4) preprocess and extract features from the real-time digital signals, and use the signal features as input values ​​for the EEG pattern recognition classifier to obtain output values. The EEG acquisition device includes a smart wearable brain ring device and a signal conversion device. The signal conversion device is integrated into the smart wearable brain ring device. The connection between the signal conversion device and the smart wearable brain ring device is an electrical connection. The connection between the signal conversion device and the front-end signal analysis and processing module is a Bluetooth communication connection. The cloud database is used to store different types of cognitive datasets. The front-end signal analysis and processing module loads the aforementioned cognitive datasets from the cloud database. The connection between the cloud database and the front-end signal analysis and processing module is a WIFI communication connection.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, The computer program processor executes a cognitive modulation method based on fun tasks and real-time EEG signals as described in any one of claims 1-4.

Images