Brain activity mode detection method and device based on multivariate multi-scale entropy

Abstract

The invention relates to a brain activity mode detection method based on multivariate and multi-scale entropy. The method comprises the following steps that electroencephalogram signals of a detected person in the game process are collected in real time; the collected electroencephalogram signals are subjected to denoising and filtering processing; carrying out complexity analysis on the preprocessed electroencephalogram signals by using a multivariate multi-scale entropy algorithm; constructing a brain activity mode feature library; and detecting and classifying the brain activity pattern by using a machine learning algorithm. The problem that in the prior art, frequency domain analysis and time domain analysis are difficult to reflect multi-dimensional and nonlinear characteristics of brain activities is solved.

Description

Technical Field

[0001] The present invention relates to the field of neuroscience technology, and in particular to a method and device for detecting brain activity patterns based on multivariate and multiscale entropy. Background Art

[0002] In the study of neuroscience and cognitive psychology, gaming behavior, as a complex decision-making process, involves multi-dimensional brain activity. Although traditional EEG analysis methods, such as frequency domain analysis and time domain analysis, can reflect certain characteristics of brain activity to some extent, they are obviously insufficient in capturing the nonlinearity, dynamics, and complexity of brain activity. Frequency domain analysis methods mainly focus on the frequency characteristics of the signal, while time domain analysis methods focus on the temporal characteristics of the signal. Both methods are difficult to reflect the multi-dimensional and nonlinear characteristics of brain activity.

[0003] Multivariate multiscale entropy (mMSE) is an algorithm that can effectively quantify the complexity of nonlinear signals, and is particularly suitable for analyzing multidimensional, nonlinear EEG signals. By introducing the mMSE method, the present invention aims to break through the limitations of traditional analysis methods and provide a more accurate and comprehensive technical means to reveal the inherent laws and potential patterns of brain activity during the game. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method and device for detecting brain activity patterns based on multivariate and multi-scale entropy, so as to solve the problem in the prior art that both frequency domain analysis and time domain analysis are difficult to reflect the multidimensional and nonlinear characteristics of brain activity.

[0005] The present invention is achieved through the following technical solutions:

[0006] A method for detecting brain activity patterns based on multivariate multiscale entropy, comprising the following steps:

[0007] S1, real-time collection of the subject's EEG signals during the game;

[0008] S2, denoising and filtering the collected EEG signals;

[0009] S3, use multivariate multiscale entropy algorithm to perform complexity analysis on the preprocessed EEG signals;

[0010] S4, build a feature library of brain activity patterns;

[0011] S5. Use machine learning algorithms to detect and classify brain activity patterns.

[0012] Furthermore, the EEG signal acquisition step uses a high-precision EEG cap with a sampling frequency of not less than 1000 Hz.

[0013] Further, the denoising and filtering process includes wavelet transform denoising and band-pass filtering, and the frequency range of the band-pass filtering is 0.5 Hz - 50 Hz.

[0014] Further, the multivariate multi-scale entropy algorithm includes the following steps:

[0015] Divide the EEG signal into multiple time windows;

[0016] Calculate the multivariate multi-scale entropy value within each time window;

[0017] Capture the non-linear characteristics of the signal through multi-scale analysis.

[0018] Further, the feature library construction step includes storing the multivariate multi-scale entropy values in different game scenarios, and the game scenarios include the prisoner's dilemma and rock-paper-scissors.

[0019] Further, the machine learning algorithm is random forest.

[0020] An apparatus applied to the above-mentioned method for detecting brain activity patterns based on multivariate multi-scale entropy includes:

[0021] An EEG signal acquisition module for real-time acquisition of the EEG signal of the subject during the game;

[0022] A signal preprocessing module for denoising and filtering the acquired EEG signal;

[0023] A multivariate multi-scale entropy calculation module for complexity analysis of the preprocessed EEG signal;

[0024] A feature library construction module for constructing a feature library of brain activity patterns;

[0025] A pattern detection module for detecting and classifying brain activity patterns using a machine learning algorithm.

[0026] Further, the EEG signal acquisition module uses a high-precision EEG cap, and the sampling frequency is not less than 1000 Hz.

[0027] Further, the signal preprocessing module includes a denoising unit and a filtering unit. The denoising unit uses wavelet transform to remove high-frequency noise, and the filtering unit uses a band-pass filter to extract EEG signals of 0.5 Hz - 50 Hz.

[0028] Further, the pattern detection module uses the random forest algorithm to compare the real-time acquired EEG signal with the data in the feature library to achieve detection and classification of brain activity patterns.

[0029] The beneficial effects of the present invention are as follows:

[0030] The method and device for detecting brain activity patterns based on multivariate multi-scale entropy can quantify the complexity of brain activity from the perspective of non-linear dynamics by introducing the multivariate multi-scale entropy algorithm, breaking through the limitations of traditional electroencephalogram (EEG) signal analysis methods.

[0031] Other advantages, objectives, and features of the present invention will, to some extent, be described in the subsequent specification, and to some extent, will be obvious to those skilled in the art based on the study of the following text, or can be learned from the practice of the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the following specification. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a schematic diagram of an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0033] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. The components of the embodiments of the present invention usually described and illustrated in the accompanying drawings here can be arranged and designed in various different configurations.

[0034] Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the present invention claimed, but merely represents selected embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0035] It should be noted that similar reference numerals and letters indicate similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings.

[0036] In the above description of the present invention, it should be noted that the orientation or positional relationship indicated by terms such as "one side" and "the other side" is based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the present invention is usually placed during use. It is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. In addition, terms such as "first" and "second" are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0037] In addition, terms such as "identical" do not require the components to be absolutely identical, but there can be minor differences. The term "vertical" only means that the positional relationship between components is more vertical relative to "parallel", and does not mean that the structure must be completely vertical, but can be slightly inclined.

[0038] Please refer to Figure 1 , the present invention provides a technical solution: a method for detecting brain activity patterns based on multivariate multi-scale entropy, comprising the following steps:

[0039] S1. Real-time collect the electroencephalogram (EEG) signals of the subject during the game process;

[0040] S2. Denoise and filter the collected EEG signals;

[0041] S3. Use the multivariate multi-scale entropy algorithm to analyze the complexity of the preprocessed EEG signals;

[0042] S4. Construct a brain activity pattern feature library;

[0043] S5. Use machine learning algorithms to detect and classify brain activity patterns.

[0044] In this solution, by introducing the multivariate multi-scale entropy algorithm, it is possible to quantify the complexity of brain activity from the perspective of nonlinear dynamics, breaking through the limitations of traditional EEG signal analysis methods.

[0045] It can simultaneously capture the subtle changes in EEG signals in time and space, comprehensively reflecting the complexity and dynamics of brain activity. Through multi-scale analysis, it is possible to accurately quantify the complexity of brain activity and reveal the internal laws of brain activity during the game process. Combining machine learning algorithms can achieve rapid detection and classification of brain activity patterns, providing strong technical support for neuroscience and cognitive psychology research.

[0046] In this embodiment: the EEG signal acquisition step uses a high-precision EEG cap with a sampling frequency of not less than 1000 Hz.

[0047] In this solution, a high-precision EEG cap (EEG cap) is used to collect the EEG signals of the subject during the game task. The signal sampling frequency is set to 1000 Hz to ensure that the spatio-temporal resolution of the signal meets the requirements of subsequent analysis.

[0048] Among them, EEG signals are typical non-stationary signals, with a fast change speed and instantaneousness. A higher sampling frequency (such as 1000 Hz) can capture the high-frequency components and rapid changes in the signal, thereby providing higher time resolution. This is crucial for analyzing the dynamic characteristics of brain activity (such as event-related potential ERP or fast neural oscillations).

[0049] In this embodiment: The denoising and filtering process includes wavelet transform denoising and band-pass filtering, and the frequency range of the band-pass filtering is 0.5 Hz - 50 Hz.

[0050] In this solution, the collected EEG signals are denoised. High-frequency noise is removed using wavelet transform, and effective EEG signals are extracted through band-pass filtering (0.5 Hz - 50 Hz).

[0051] Among them, EEG signals usually contain low-frequency drifts (such as baseline fluctuations caused by breathing, sweating, or electrode movement), and the frequencies of these components are usually lower than 0.5 Hz. By setting high-pass filtering (0.5 Hz), these low-frequency interferences can be effectively removed. High-frequency noise in EEG signals (such as electromyogram signals, power line interference, etc.) is usually concentrated above 50 Hz. By setting low-pass filtering (50 Hz), these high-frequency interferences can be effectively removed.

[0052] Furthermore, the effective frequency range of EEG signals (0.5 Hz - 50 Hz) is retained, covering the main frequency bands related to cognition and neural activities, such as:

[0053] Delta wave (0.5 Hz - 4 Hz): Related to deep sleep and recovery processes;

[0054] Theta wave (4 Hz - 8 Hz): Related to memory, emotion, and attention;

[0055] Alpha wave (8 Hz - 13 Hz): Related to relaxation and resting state;

[0056] Beta wave (13 Hz - 30 Hz): Related to attention, thinking, and decision-making;

[0057] Gamma wave (30 Hz - 50 Hz): Related to higher cognitive functions and neural synchronization.

[0058] In this embodiment: The multivariate multiscale entropy algorithm includes the following steps:

[0059] Divide the EEG signal into multiple time windows;

[0060] Calculate the multivariate multiscale entropy value within each time window;

[0061] Capture the non-linear characteristics of the signal through multiscale analysis.

[0062] In this solution, the multivariate multiscale entropy (mMSE) algorithm is used to analyze the preprocessed EEG signals. The specific steps are as follows:

[0063] Step 1: Divide the preprocessed EEG signal into multiple time windows in chronological order. The length of each window is 1 second. For example, for a signal with a sampling frequency of 1000 Hz, each time window contains 1000 data points. By dividing the time windows, the continuous EEG signal can be converted into discrete time periods, which is convenient for subsequent multi-scale analysis and complexity calculation.

[0064] Step 2: For the EEG signal in each time window, the multivariate multiscale entropy (mMSE) algorithm is used to calculate its entropy value. Multi-channel EEG signals (such as data from different electrodes) are integrated into multivariate signals to reflect the spatial distribution characteristics of brain activity, and the multivariate signals are coarse-grained to generate signal sequences at multiple time scales. The coarse-graining process is achieved by calculating the average value at each scale. For the coarse-grained signal sequence, the multivariate sample entropy (MSE) is calculated. Multivariate sample entropy is an indicator used to quantify the complexity of multivariate signals and can reflect the nonlinear characteristics of the signal. By calculating the multivariate multiscale entropy value, the complexity of EEG signals at different time scales can be quantified, reflecting the nonlinear dynamic characteristics of brain activity.

[0065] Step 3: Analyze the multivariate and multiscale entropy changes of EEG signals at multiple time scales to capture the nonlinear characteristics of the signal. Select multiple time scales (such as scale 1 to scale 20) to cover different time ranges from short to long. Analyze the entropy changes at each time scale and observe the trend of signal complexity changing with scale. Extract the entropy features at multiple scales and construct a feature vector that reflects the complexity of brain activity. Through multi-scale analysis, the nonlinear characteristics of EEG signals at different time scales can be fully reflected, revealing the inherent laws of brain activity.

[0066] Step 4: Integrate the multi-scale entropy values ​​of each time window into a feature vector and construct a brain activity pattern feature library. The feature library contains multi-scale entropy value features under different game scenarios, providing a systematic reference for subsequent pattern detection and classification.

[0067] In this embodiment: the feature library construction step includes storing multivariate and multiscale entropy values ​​under different game scenarios, and the game scenarios include prisoner's dilemma and rock-paper-scissors.

[0068] In this scheme, the EEG signals of the subjects are collected in different game scenarios (such as prisoner's dilemma, rock-paper-scissors, ultimatum game, etc.), and the multivariate multiscale entropy (mMSE) algorithm is used to calculate its complexity characteristics, providing the original data and label information for the feature library to ensure that the data in the feature library has clear classification and traceability.

[0069] The data for each gaming scenario includes:

[0070] 1. EEG signal data: multi-channel EEG signals, covering activity information of different brain regions.

[0071] 2. mMSE value: The multivariate multiscale entropy value in each time window reflects the complexity of the signal at multiple scales.

[0072] 3. Game scenario label: mark the game task type corresponding to each data sample (such as prisoner's dilemma, rock-paper-scissors, etc.).

[0073] In this embodiment: the machine learning algorithm is random forest.

[0074] In this solution, the real-time collected EEG signals are compared with the data in the feature library and classified using the trained machine learning model. This enables real-time detection and classification of brain activity patterns, providing real-time technical support for neuroscience research and practical applications.

[0075] The specific steps include:

[0076] Signal preprocessing: Perform preprocessing operations such as denoising and filtering on the real-time collected EEG signals.

[0077] Feature extraction: The multivariate multiscale entropy (mMSE) algorithm is used to extract the complexity characteristics of the signal.

[0078] Pattern detection: The extracted features are fed into a trained machine learning model for pattern detection and classification.

[0079] Result output: Output classification results and identify the game situation corresponding to the current brain activity pattern.

[0080] Among them, a variety of machine learning algorithms can be used for pattern detection and classification, preferably but not limited to random forest, and can also be support vector machine (SVM), neural network (Neura l Networks), gradient boosting tree (GBDT) and K nearest neighbor algorithm (KNN), etc. According to specific needs, the appropriate machine learning algorithm is selected to improve the accuracy and efficiency of pattern detection and classification, and provide diversified technical support for the detection and classification of brain activity patterns.

[0081] A device applied to the above-mentioned brain activity pattern detection method based on multivariate multiscale entropy, comprising:

[0082] The EEG signal acquisition module is used to collect the EEG signals of the subject in real time during the game;

[0083] A signal preprocessing module is used to perform denoising and filtering on the collected EEG signals;

[0084] A multi - variable multi - scale entropy calculation module for performing complexity analysis on the pre - processed EEG signals;

[0085] A feature library construction module for constructing a feature library of brain activity patterns;

[0086] A pattern detection module for detecting and classifying brain activity patterns using machine learning algorithms.

[0087] In this solution, the EEG signal acquisition module is used to collect the EEG signals of the subject in real - time during the game process, providing raw data for subsequent analysis; the signal pre - processing module performs denoising and filtering on the collected EEG signals to remove interference signals and extract effective EEG signal components; the multi - variable multi - scale entropy calculation module performs complexity analysis on the pre - processed EEG signals to quantify the non - linear characteristics of brain activity; the feature library construction module constructs a feature library of brain activity patterns to store the complexity characteristics of EEG signals in different game scenarios; the pattern detection module uses machine learning algorithms to detect and classify brain activity patterns to identify brain activity patterns in different game scenarios.

[0088] Through modular design, the acquisition, pre - processing, complexity analysis, feature library construction, and pattern detection and classification of EEG signals are realized, which have the advantages of high precision, high efficiency, intelligence, and scalability, providing strong technical support for neuroscience and cognitive psychology research.

[0089] In this embodiment: The EEG signal acquisition module uses a high - precision EEG cap with a sampling frequency of not less than 1000Hz.

[0090] In this solution, the high - precision EEG cap uses multi - channel electrodes (such as 64 - channel or 128 - channel), covering multiple regions of the brain, ensuring the spatial resolution of the signals. The sampling frequency is not less than 1000Hz, providing high temporal resolution, capable of capturing rapid changes and high - frequency components in EEG signals, meeting the requirements of the Nyquist sampling theorem, and providing high - quality raw data for subsequent signal processing and analysis.

[0091] In this embodiment: The signal pre - processing module includes a denoising unit and a filtering unit. The denoising unit uses wavelet transform to remove high - frequency noise, and the filtering unit uses a band - pass filter to extract EEG signals in the range of 0.5Hz - 50Hz.

[0092] In this solution, the denoising unit uses wavelet transform to remove high-frequency noise (such as electromyogram signals, power line interference, etc.) and retains the effective electroencephalogram signal components. The filtering unit uses a band-pass filter to extract electroencephalogram signals in the range of 0.5 Hz - 50 Hz, removes low-frequency drift (such as baseline fluctuations) and high-frequency noise, and retains the effective frequency band related to brain activities. The signal-to-noise ratio (SNR) is improved to provide clean input data for subsequent complexity analysis and pattern detection.

[0093] In this embodiment: The pattern detection module uses the random forest algorithm to compare the real-time collected electroencephalogram signals with the data in the feature library to achieve the detection and classification of brain activity patterns.

[0094] In this solution, first, the random forest model is trained using the data in the feature library to optimize the model parameters; then, the real-time collected electroencephalogram signals are compared with the data in the feature library for classification; finally, the classification results are output to identify the game scenario corresponding to the current brain activity pattern. The accurate detection and classification of brain activity patterns are achieved, providing technical support for neuroscience research and practical applications.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the purpose and scope of the technical solutions of the present invention, and they should all be covered within the scope of the claims of the present invention.

Claims

1. A method for detecting brain activity patterns based on multivariate multi-scale entropy, characterized in that: It includes the following steps: S1. Real-time collect the electroencephalogram (EEG) signals of the subject during the game; S2. Denoise and filter the collected EEG signals; S3. Use the multivariate multi-scale entropy algorithm to analyze the complexity of the preprocessed EEG signals; S4. Construct a feature library of brain activity patterns; S5. Use machine learning algorithms to detect and classify brain activity patterns.

2. The method for detecting brain activity patterns based on multivariate multi-scale entropy according to claim 1, characterized in that: The EEG signal collection step uses a high-precision EEG cap, and the sampling frequency is not less than 1000Hz.

3. The method for detecting brain activity patterns based on multivariate multi-scale entropy according to claim 1, wherein: The denoising and filtering process includes wavelet transform denoising and band-pass filtering, and the frequency range of the band-pass filtering is 0.5Hz - 50Hz.

4. The method for detecting brain activity patterns based on multivariate multi-scale entropy according to claim 1, wherein: The multivariate multi-scale entropy algorithm includes the following steps: Divide the EEG signals into multiple time windows; Calculate the multivariate multi-scale entropy values within each time window; Capture the non-linear features of the signals through multi-scale analysis.

5. The method for detecting brain activity patterns based on multivariate multi-scale entropy according to claim 1, wherein: The feature library construction step includes storing the multivariate multi-scale entropy values under different game scenarios, and the game scenarios include the prisoner's dilemma and rock-paper-scissors.

6. The method for detecting brain activity patterns based on multivariate multi-scale entropy according to claim 1, characterized in that: The machine learning algorithm is the random forest.

7. An apparatus for the method for detecting brain activity patterns based on multivariate multi-scale entropy according to claim 1, characterized in that: It includes: An EEG signal collection module, which is used to real-time collect the EEG signals of the subject during the game; A signal preprocessing module, which is used to denoise and filter the collected EEG signals; A multivariate multi-scale entropy calculation module, which is used to analyze the complexity of the preprocessed EEG signals; A feature library construction module, which is used to construct a feature library of brain activity patterns; A pattern detection module, which is used to detect and classify brain activity patterns by using machine learning algorithms.

8. The device according to claim 7, characterized in that: The EEG signal collection module uses a high-precision EEG cap, and the sampling frequency is not less than 1000Hz.

9. The device according to claim 7, characterized in that: The signal preprocessing module includes a denoising unit and a filtering unit. The denoising unit uses wavelet transform to remove high-frequency noise, and the filtering unit uses a band-pass filter to extract EEG signals of 0.5Hz - 50Hz.

10. The device according to claim 7, characterized in that: The pattern detection module uses the random forest algorithm to compare the real-time collected EEG signals with the data in the feature library to achieve the detection and classification of brain activity patterns.

Images